title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Hierarchical robust aggregation of sales forecasts at aggregated levels
in e-commerce, based on exponential smoothing and Holt's linear trend method | We revisit the interest of classical statistical techniques for sales
forecasting like exponential smoothing and extensions thereof (as Holt's linear
trend method). We do so by considering ensemble forecasts, given by several
instances of these classical techniques tuned with different (sets of)
parameters, and by forming convex combinations of the elements of ensemble
forecasts over time, in a robust and sequential manner. The machine-learning
theory behind this is called "robust online aggregation", or "prediction with
expert advice", or "prediction of individual sequences" (see Cesa-Bianchi and
Lugosi, 2006). We apply this methodology to a hierarchical data set of sales
provided by the e-commerce company Cdiscount and output forecasts at the levels
of subsubfamilies, subfamilies and families of items sold, for various
forecasting horizons (up to 6-week-ahead). The performance achieved is better
than what would be obtained by optimally tuning the classical techniques on a
train set and using their forecasts on the test set. The performance is also
good from an intrinsic point of view (in terms of mean absolute percentage of
error). While getting these better forecasts of sales at the levels of
subsubfamilies, subfamilies and families is interesting per se, we also suggest
to use them as additional features when forecasting demand at the item level.
| stat |
Short-term photovoltaic generation forecasting using multiple
heterogenous sources of data | Renewable Energies (RES) penetration is progressing rapidly: in France, the
installed capacity of photovoltaic (PV) power rose from 26MW in 2007 to 8GW in
2017 [1]. Power generated by PV plants being highly dependent on variable
weather conditions, this ever-growing pace is raising issues regarding grid
stability and revenue optimization. To overcome these obstacles, PV forecasting
became an area of intense research. In this paper, we propose a low complexity
forecasting model able to operate with multiple heterogenous sources of data
(power measurements, satellite images and Numerical Weather Predictions (NWP)).
Being non-parametric, this model can be extended to include inputs. The main
strength of the proposed model lies in its ability to automatically select the
optimal sources of data according to the desired forecast horizon (from 15min
to 6h ahead) thanks to a feature selection procedure. To take advantage of the
growing number of PV plants, a Spatio-Temporal (ST) approach is implemented.
This approach considers the dependencies between spatially distributed plants.
Each source has been studied incrementally so as to quantify their impact on
forecast performances. This plurality of sources enhances the forecasting
performances up to 40% in terms of RMSE compared to a reference model. The
evaluation process is carried out on nine PV plants from the Compagnie
Nationale du Rh{\^o}ne (CNR).
| stat |
Inference tools for Markov Random Fields on lattices: The R package
mrf2d | Markov random fields on two-dimensional lattices are behind many image
analysis methodologies. mrf2d provides tools for statistical inference on a
class of discrete stationary Markov random field models with pairwise
interaction, which includes many of the popular models such as the Potts model
and texture image models. The package introduces representations of dependence
structures and parameters, visualization functions and efficient (C++ based)
implementations of sampling algorithms, common estimation methods and other key
features of the model, providing a useful framework to implement algorithms and
working with the model in general. This paper presents a description and
details of the package, as well as some reproducible examples of usage.
| stat |
On function-on-function regression: Partial least squares approach | Functional data analysis tools, such as function-on-function regression
models, have received considerable attention in various scientific fields
because of their observed high-dimensional and complex data structures. Several
statistical procedures, including least squares, maximum likelihood, and
maximum penalized likelihood, have been proposed to estimate such
function-on-function regression models. However, these estimation techniques
produce unstable estimates in the case of degenerate functional data or are
computationally intensive. To overcome these issues, we proposed a partial
least squares approach to estimate the model parameters in the
function-on-function regression model. In the proposed method, the B-spline
basis functions are utilized to convert discretely observed data into their
functional forms. Generalized cross-validation is used to control the degrees
of roughness. The finite-sample performance of the proposed method was
evaluated using several Monte-Carlo simulations and an empirical data analysis.
The results reveal that the proposed method competes favorably with existing
estimation techniques and some other available function-on-function regression
models, with significantly shorter computational time.
| stat |
Utility/Privacy Trade-off through the lens of Optimal Transport | Strategic information is valuable either by remaining private (for instance
if it is sensitive) or, on the other hand, by being used publicly to increase
some utility. These two objectives are antagonistic and leaking this
information might be more rewarding than concealing it. Unlike classical
solutions that focus on the first point, we consider instead agents that
optimize a natural trade-off between both objectives. We formalize this as an
optimization problem where the objective mapping is regularized by the amount
of information revealed to the adversary (measured as a divergence between the
prior and posterior on the private knowledge). Quite surprisingly, when
combined with the entropic regularization, the Sinkhorn loss naturally emerges
in the optimization objective, making it efficiently solvable. We apply these
techniques to preserve some privacy in online repeated auctions.
| stat |
Causal identification of infectious disease intervention effects in a
clustered population | Causal identification of treatment effects for infectious disease outcomes in
interconnected populations is challenging because infection outcomes may be
transmissible to others, and treatment given to one individual may affect
others' outcomes. Contagion, or transmissibility of outcomes, complicates
standard conceptions of treatment interference in which an intervention
delivered to one individual can affect outcomes of others. Several statistical
frameworks have been proposed to measure causal treatment effects in this
setting, including structural transmission models, mediation-based partnership
models, and randomized trial designs. However, existing estimands for
infectious disease intervention effects are of limited conceptual usefulness:
Some are parameters in a structural model whose causal interpretation is
unclear, others are causal effects defined only in a restricted two-person
setting, and still others are nonparametric estimands that arise naturally in
the context of a randomized trial but may not measure any biologically
meaningful effect. In this paper, we describe a unifying formalism for defining
nonparametric structural causal estimands and an identification strategy for
learning about infectious disease intervention effects in clusters of
interacting individuals when infection times are observed. The estimands
generalize existing quantities and provide a framework for causal
identification in randomized and observational studies, including situations
where only binary infection outcomes are observed. A semiparametric class of
pairwise Cox-type transmission hazard models is used to facilitate statistical
inference in finite samples. A comprehensive simulation study compares existing
and proposed estimands under a variety of randomized and observational vaccine
trial designs.
| stat |
Impact of Narrow Lanes on Arterial Road Vehicle Crashes: A Machine
Learning Approach | In this paper we adopted state-of-the-art machine learning algorithms,
namely: random forest (RF) and least squares boosting, to model crash data and
identify the optimum model to study the impact of narrow lanes on the safety of
arterial roads. Using a ten-year crash dataset in four cities in Nebraska, two
machine learning models were assessed based on the prediction error. The RF
model was identified as the best model. The RF was used to compute the
importance of the lane width predictors in our regression model based on two
different measures. Subsequently, the RF model was used to simulate the crash
rate for different lane widths. The Kruskal-Wallis test, was then conducted to
determine if simulated values from the four lane width groups have equal means.
The test null hypothesis of equal means for simulated values from the four lane
width groups was rejected. Consequently, it was concluded that the crash rates
from at least one lane width group was statistically different from the others.
Finally, the results from the pairwise comparisons using the Tukey and Kramer
test showed that the changes in crash rates between any two lane width
conditions were statistically significant.
| stat |
A Random Forest Approach for Modeling Bounded Outcomes | Random forests have become an established tool for classification and
regression, in particular in high-dimensional settings and in the presence of
complex predictor-response relationships. For bounded outcome variables
restricted to the unit interval, however, classical random forest approaches
may severely suffer as they do not account for the heteroscedasticity in the
data. A random forest approach is proposed for relating beta distributed
outcomes to explanatory variables. The approach explicitly makes use of the
likelihood function of the beta distribution for the selection of splits during
the tree-building procedure. In each iteration of the tree-building algorithm
one chooses the combination of explanatory variable and splitting rule that
maximizes the log-likelihood function of the beta distribution with the
parameter estimates derived from the nodes of the currently built tree. Several
simulation studies demonstrate the properties of the method and compare its
performance to classical random forest approaches as well as to parametric
regression models.
| stat |
Model Specification Test with Unlabeled Data: Approach from Covariate
Shift | We propose a novel framework of the model specification test in regression
using unlabeled test data. In many cases, we have conducted statistical
inferences based on the assumption that we can correctly specify a model.
However, it is difficult to confirm whether a model is correctly specified. To
overcome this problem, existing works have devised statistical tests for model
specification. Existing works have defined a correctly specified model in
regression as a model with zero conditional mean of the error term over train
data only. Extending the definition in conventional statistical tests, we
define a correctly specified model as a model with zero conditional mean of the
error term over any distribution of the explanatory variable. This definition
is a natural consequence of the orthogonality of the explanatory variable and
the error term. If a model does not satisfy this condition, the model might
lack robustness with regards to the distribution shift. The proposed method
would enable us to reject a misspecified model under our definition. By
applying the proposed method, we can obtain a model that predicts the label for
the unlabeled test data well without losing the interpretability of the model.
In experiments, we show how the proposed method works for synthetic and
real-world datasets.
| stat |
Hypothesis-based acceptance sampling for modules F and F1 of the
European Measuring Instruments Directive | Millions of measuring instruments are verified each year before being placed
on the markets worldwide. In the EU, such initial conformity assessments are
regulated by the Measuring Instruments Directive (MID). The MID modules F and
F1 on product verification allow for statistical acceptance sampling, whereby
only random subsets of instruments need to be inspected. This paper
re-interprets the acceptance sampling conditions formulated by the MID. The new
interpretation is contrasted with the one advanced in WELMEC guide 8.10, and
three advantages have become apparent. Firstly, an economic advantage of the
new interpretation is a producers' risk bounded from above, such that measuring
instruments with sufficient quality are accepted with a guaranteed probability
of no less than 95 %. Secondly, a conceptual advantage is that the new MID
interpretation fits into the well-known, formal framework of statistical
hypothesis testing. Thirdly, the new interpretation applies unambiguously to
finite-sized lots, even very small ones. We conclude that the new
interpretation is to be preferred and suggest re-formulating the statistical
sampling conditions in the MID. Re-interpreting the MID conditions implies that
currently available sampling plans are either not admissible or not optimal. We
derive a new acceptance sampling scheme and recommend its application.
| stat |
A Bayesian Approach to Joint Estimation of Multiple Graphical Models | The problem of joint estimation of multiple graphical models from high
dimensional data has been studied in the statistics and machine learning
literature, due to its importance in diverse fields including molecular
biology, neuroscience and the social sciences. This work develops a Bayesian
approach that decomposes the model parameters across the multiple graphical
models into shared components across subsets of models and edges, and
idiosyncratic ones. Further, it leverages a novel multivariate prior
distribution, coupled with a pseudo-likelihood that enables fast computations
through a robust and efficient Gibbs sampling scheme. We establish strong
posterior consistency for model selection, as well as estimation of model
parameters under high dimensional scaling with the number of variables growing
exponentially with the sample size. The efficacy of the proposed approach is
illustrated on both synthetic and real data.
Keywords: Pseudo-likelihood, Gibbs sampling, posterior consistency, Omics
data
| stat |
Hyperbolic Manifold Regression | Geometric representation learning has recently shown great promise in several
machine learning settings, ranging from relational learning to language
processing and generative models. In this work, we consider the problem of
performing manifold-valued regression onto an hyperbolic space as an
intermediate component for a number of relevant machine learning applications.
In particular, by formulating the problem of predicting nodes of a tree as a
manifold regression task in the hyperbolic space, we propose a novel
perspective on two challenging tasks: 1) hierarchical classification via label
embeddings and 2) taxonomy extension of hyperbolic representations. To address
the regression problem we consider previous methods as well as proposing two
novel approaches that are computationally more advantageous: a parametric deep
learning model that is informed by the geodesics of the target space and a
non-parametric kernel-method for which we also prove excess risk bounds. Our
experiments show that the strategy of leveraging the hyperbolic geometry is
promising. In particular, in the taxonomy expansion setting, we find that the
hyperbolic-based estimators significantly outperform methods performing
regression in the ambient Euclidean space.
| stat |
Is the new model better? One metric says yes, but the other says no.
Which metric do I use? | Incremental value (IncV) evaluates the performance change from an existing
risk model to a new model. It is one of the key considerations in deciding
whether a new risk model performs better than the existing one. Problems arise
when different IncV metrics contradict each other. For example, compared with a
prescribed-dose model, an ovarian-dose model for predicting acute ovarian
failure has a slightly lower area under the receiver operating characteristic
curve (AUC) but increases the area under the precision-recall curve (AP) by
48%. This phenomenon of conflicting conclusions is not uncommon, and it creates
a dilemma in medical decision making. In this article, we examine the
analytical connections and differences between two IncV metrics: IncV in AUC
(IncV-AUC) and IncV in AP (IncV-AP). Additionally, since they are both
semi-proper scoring rules, we compare them with a strictly proper scoring rule:
the IncV of the scaled Brier score (IncV-sBrS), via a numerical study. We
demonstrate that both IncV-AUC and IncV-AP are weighted averages of the changes
(from the existing model to the new one) in separating the risk score
distributions between events and non-events. However, IncV-AP assigns heavier
weights to the changes in the high-risk group, whereas IncV-AUC weights the
changes equally. In the numerical study, we find that IncV-AP has a wide range,
from negative to positive, but the size of IncV-AUC is much smaller. In
addition, IncV-AP and IncV-sBr Sare highly consistent, but IncV-AUC is
negatively correlated with IncV-sBrS and IncV-AP at a low event rate. IncV-AUC
and IncV-AP are the least consistent among the three pairs, and their
differences are more pronounced as the event rate decreases.
| stat |
Fitting Prediction Rule Ensembles with R Package pre | Prediction rule ensembles (PREs) are sparse collections of rules, offering
highly interpretable regression and classification models. This paper presents
the R package pre, which derives PREs through the methodology of Friedman and
Popescu (2008). The implementation and functionality of package pre is
described and illustrated through application on a dataset on the prediction of
depression. Furthermore, accuracy and sparsity of PREs is compared with that of
single trees, random forest and lasso regression in four benchmark datasets.
Results indicate that pre derives ensembles with predictive accuracy comparable
to that of random forests, while using a smaller number of variables for
prediction.
| stat |
Central Moment Discrepancy (CMD) for Domain-Invariant Representation
Learning | The learning of domain-invariant representations in the context of domain
adaptation with neural networks is considered. We propose a new regularization
method that minimizes the discrepancy between domain-specific latent feature
representations directly in the hidden activation space. Although some standard
distribution matching approaches exist that can be interpreted as the matching
of weighted sums of moments, e.g. Maximum Mean Discrepancy (MMD), an explicit
order-wise matching of higher order moments has not been considered before. We
propose to match the higher order central moments of probability distributions
by means of order-wise moment differences. Our model does not require
computationally expensive distance and kernel matrix computations. We utilize
the equivalent representation of probability distributions by moment sequences
to define a new distance function, called Central Moment Discrepancy (CMD). We
prove that CMD is a metric on the set of probability distributions on a compact
interval. We further prove that convergence of probability distributions on
compact intervals w.r.t. the new metric implies convergence in distribution of
the respective random variables. We test our approach on two different
benchmark data sets for object recognition (Office) and sentiment analysis of
product reviews (Amazon reviews). CMD achieves a new state-of-the-art
performance on most domain adaptation tasks of Office and outperforms networks
trained with MMD, Variational Fair Autoencoders and Domain Adversarial Neural
Networks on Amazon reviews. In addition, a post-hoc parameter sensitivity
analysis shows that the new approach is stable w.r.t. parameter changes in a
certain interval. The source code of the experiments is publicly available.
| stat |
Spectral Algorithms for Community Detection in Directed Networks | Community detection in large social networks is affected by degree
heterogeneity of nodes. The D-SCORE algorithm for directed networks was
introduced to reduce this effect by taking the element-wise ratios of the
singular vectors of the adjacency matrix before clustering. Meaningful results
were obtained for the statistician citation network, but rigorous analysis on
its performance was missing. First, this paper establishes theoretical
guarantee for this algorithm and its variants for the directed degree-corrected
block model (Directed-DCBM). Second, this paper provides significant
improvements for the original D-SCORE algorithms by attaching the nodes outside
of the community cores using the information of the original network instead of
the singular vectors.
| stat |
Fr\'echet Change Point Detection | We propose a method to infer the presence and location of change-points in
the distribution of a sequence of independent data taking values in a general
metric space, where change-points are viewed as locations at which the
distribution of the data sequence changes abruptly in terms of either its
Fr\'echet mean or Fr\'echet variance or both. The proposed method is based on
comparisons of Fr\'echet variances before and after putative change-point
locations and does not require a tuning parameter except for the specification
of cut-off intervals near the endpoints where change-points are assumed not to
occur. Our results include theoretical guarantees for consistency of the test
under contiguous alternatives when a change-point exists and also for
consistency of the estimated location of the change-point if it exists, where
under the null hypothesis of no change-point the limit distribution of the
proposed scan function is the square of a standardized Brownian Bridge. These
consistency results are applicable for a broad class of metric spaces under
mild entropy conditions. Examples include the space of univariate probability
distributions and the space of graph Laplacians for networks. Simulation
studies demonstrate the effectiveness of the proposed methods, both for
inferring the presence of a change-point and estimating its location. We also
develop theory that justifies bootstrap-based inference and illustrate the new
approach with sequences of maternal fertility distributions and communication
networks.
| stat |
Tutorial: Introduction to computational causal inference using
reproducible Stata, R and Python code | The purpose of many health studies is to estimate the effect of an exposure
on an outcome. It is not always ethical to assign an exposure to individuals in
randomised controlled trials, instead observational data and appropriate study
design must be used. There are major challenges with observational studies, one
of which is confounding that can lead to biased estimates of the causal
effects. Controlling for confounding is commonly performed by simple adjustment
for measured confounders; although, often this is not enough. Recent advances
in the field of causal inference have dealt with confounding by building on
classical standardisation methods. However, these recent advances have
progressed quickly with a relative paucity of computational-oriented applied
tutorials contributing to some confusion in the use of these methods among
applied researchers. In this tutorial, we show the computational implementation
of different causal inference estimators from a historical perspective where
different estimators were developed to overcome the limitations of the previous
one. Furthermore, we also briefly introduce the potential outcomes framework,
illustrate the use of different methods using an illustration from the health
care setting, and most importantly, we provide reproducible and commented code
in Stata, R and Python for researchers to apply in their own observational
study. The code can be accessed at
https://github.com/migariane/TutorialCausalInferenceEstimators
| stat |
Covariance-based Dissimilarity Measures Applied to Clustering Wide-sense
Stationary Ergodic Processes | We introduce a new unsupervised learning problem: clustering wide-sense
stationary ergodic stochastic processes. A covariance-based dissimilarity
measure together with asymptotically consistent algorithms is designed for
clustering offline and online datasets, respectively. We also suggest a formal
criterion on the efficiency of dissimilarity measures, and discuss of some
approach to improve the efficiency of our clustering algorithms, when they are
applied to cluster particular type of processes, such as self-similar processes
with wide-sense stationary ergodic increments. Clustering synthetic data and
real-world data are provided as examples of applications.
| stat |
Comprehensive cluster validity Index based on structural simplicity | Nonhierarchical clustering depending on unsupervised algorithms may not
retrieve the optimal partition of datasets. Determining if clusters fit
``natural partitions`` can be achieved using cluster validity indices (CVIs).
Most existing CVIs consider criteria such as cohesion, separation, and their
equivalents. However, these binary relations may provide neither the optimal
measure of partition suitability nor reference values corresponding to the
worst partition. Moreover, previous CVI studies have been mostly focused on
fitting correct partitions according to researchers' a priori assumptions. In
contrast, we investigated desirable properties of CVIs, namely, scale shift
transform invariance, optimal clustering, and unbiased clustering with
representing the worst partition. Then, we conducted experiments to evaluate
whether existing CVIs fulfill these properties. As none of these CVIs fulfilled
the desired properties, we propose the simplicity index, which measures the
simplicity of tree structures in clusters. The simplicity index is the unique
index invariant to the ``correct rate`` and provides both a reference
indicating the most complex partition and the best value indicating the
simplest one.
| stat |
Analyzing Cause-Specific Mortality Trends using Compositional Functional
Data Analysis | We study the dynamics of cause--specific mortality rates among countries by
considering them as compositions of functions. We develop a novel framework for
such data structure, with particular attention to functional PCA. The
application of this method to a subset of the WHO mortality database reveals
the main modes of variation of cause--specific rates over years for men and
women and enables us to perform clustering in the projected subspace. The
results give many insights of the ongoing trends, only partially explained by
past literature, that the considered countries are undergoing. We are also able
to show the different evolution of cause of death undergone by men and women:
for example, we can see that while lung cancer incidence is stabilizing for
men, it is still increasing for women.
| stat |
Nonparametric tests for transition probabilities in nonhomogeneous
Markov processes | This paper proposes nonparametric two-sample tests for the direct comparison
of the probabilities of a particular transition between states of a continuous
time nonhomogeneous Markov process with a finite state space. The proposed
tests are a linear nonparametric test, an L2-norm-based test and a
Kolmogorov-Smirnov-type test. Significance level assessment is based on
rigorous procedures, which are justified through the use of modern empirical
process theory. Moreover, the L2-norm and the Kolmogorov-Smirnov-type tests are
shown to be consistent for every fixed alternative hypothesis. The proposed
tests are also extended to more complex situations such as cases with
incompletely observed absorbing states and non-Markov processes. Simulation
studies show that the test statistics perform well even with small sample
sizes. Finally, the proposed tests are applied to data on the treatment of
early breast cancer from the European Organization for Research and Treatment
of Cancer (EORTC) trial 10854, under an illness-death model.
| stat |
Interpretable pathological test for Cardio-vascular disease: Approximate
Bayesian computation with distance learning | Cardio/cerebrovascular diseases (CVD) have become one of the major health
issue in our societies. But recent studies show that the present clinical tests
to detect CVD are ineffectual as they do not consider different stages of
platelet activation or the molecular dynamics involved in platelet interactions
and are incapable to consider inter-individual variability. Here we propose a
stochastic platelet deposition model and an inferential scheme for uncertainty
quantification of these parameters using Approximate Bayesian Computation and
distance learning. Finally we show that our methodology can learn biologically
meaningful parameters, which are the specific dysfunctioning parameters in each
type of patients, from data collected from healthy volunteers and patients.
This work opens up an unprecedented opportunity of personalized pathological
test for CVD detection and medical treatment. Also our proposed methodology can
be used to other fields of science where we would need machine learning tools
to be interpretable.
| stat |
Tree-Sliced Variants of Wasserstein Distances | Optimal transport (\OT) theory defines a powerful set of tools to compare
probability distributions. \OT~suffers however from a few drawbacks,
computational and statistical, which have encouraged the proposal of several
regularized variants of OT in the recent literature, one of the most notable
being the \textit{sliced} formulation, which exploits the closed-form formula
between univariate distributions by projecting high-dimensional measures onto
random lines. We consider in this work a more general family of ground metrics,
namely \textit{tree metrics}, which also yield fast closed-form computations
and negative definite, and of which the sliced-Wasserstein distance is a
particular case (the tree is a chain). We propose the tree-sliced Wasserstein
distance, computed by averaging the Wasserstein distance between these measures
using random tree metrics, built adaptively in either low or high-dimensional
spaces. Exploiting the negative definiteness of that distance, we also propose
a positive definite kernel, and test it against other baselines on a few
benchmark tasks.
| stat |
Bias Reduction as a Remedy to the Consequences of Infinite Estimates in
Poisson and Tobit Regression | Data separation is a well-studied phenomenon that can cause problems in the
estimation and inference from binary response models. Complete or
quasi-complete separation occurs when there is a combination of regressors in
the model whose value can perfectly predict one or both outcomes. In such
cases, and such cases only, the maximum likelihood estimates and the
corresponding standard errors are infinite. It is less widely known that the
same can happen in further microeconometric models. One of the few works in the
area is Santos Silva and Tenreyro (2010) who note that the finiteness of the
maximum likelihood estimates in Poisson regression depends on the data
configuration and propose a strategy to detect and overcome the consequences of
data separation. However, their approach can lead to notable bias on the
parameter estimates when the regressors are correlated. We illustrate how
bias-reducing adjustments to the maximum likelihood score equations can
overcome the consequences of separation in Poisson and Tobit regression models.
| stat |
Borrowing of information across patient subgroups in a basket trial
based on distributional discrepancy | Basket trials have emerged as a new class of efficient approaches in oncology
to evaluate a new treatment in several patient subgroups simultaneously. In
this paper, we extend the key ideas to disease areas outside of oncology,
developing a robust Bayesian methodology for randomised, placebo-controlled
basket trials with a continuous endpoint to enable borrowing of information
across subtrials with similar treatment effects. After adjusting for
covariates, information from a complementary subtrial can be represented into a
commensurate prior for the parameter that underpins the subtrial under
consideration. We propose using distributional discrepancy to characterise the
commensurability between subtrials for appropriate borrowing of information
through a spike-and-slab prior, which is placed on the prior precision factor.
When the basket trial has at least three subtrials, commensurate priors for
point-to-point borrowing are combined into a marginal predictive prior,
according to the weights transformed from the pairwise discrepancy measures. In
this way, only information from subtrial(s) with the most commensurate
treatment effect is leveraged. The marginal predictive prior is updated to a
robust posterior by the contemporary subtrial data to inform decision making.
Operating characteristics of the proposed methodology are evaluated through
simulations motivated by a real basket trial in chronic diseases. The proposed
methodology has advantages compared to other selected Bayesian analysis models,
for (i) identifying the most commensurate source of information, and (ii)
gauging the degree of borrowing from specific subtrials. Numerical results also
suggest that our methodology can improve the precision of estimates and,
potentially, the statistical power for hypothesis testing.
| stat |
Analysis and Forecasting of Fire incidence in Davao City | Fire incidence is a big problem for every local government unit in the
Philippines. The two most detrimental effects of fire incidence are economic
loss and loss of life. To mitigate these losses, proper planning and
implementation of control measures must be done. An essential aspect of
planning and control measures is prediction of possible fire incidences. This
study is conducted to analyze the historical data to create a forecasting model
for the fire incidence in Davao City. Results of the analyses show that fire
incidence has no trend or seasonality, and occurrences of fire are neither
consistently increasing nor decreasing over time. Furthermore, the absence of
seasonality in the data indicate that surge of fire incidence may occur at any
time of the year. Therefore, fire prevention activities should be done all year
round and not just during fire prevention month.
| stat |
Non-parametric generalized linear model | A fundamental problem in statistical neuroscience is to model how neurons
encode information by analyzing electrophysiological recordings. A popular and
widely-used approach is to fit the spike trains with an autoregressive point
process model. These models are characterized by a set of convolutional
temporal filters, whose subsequent analysis can help reveal how neurons encode
stimuli, interact with each other, and process information. In practice a
sufficiently rich but small ensemble of temporal basis functions needs to be
chosen to parameterize the filters. However, obtaining a satisfactory fit often
requires burdensome model selection and fine tuning the form of the basis
functions and their temporal span. In this paper we propose a nonparametric
approach for jointly inferring the filters and hyperparameters using the
Gaussian process framework. Our method is computationally efficient taking
advantage of the sparse variational approximation while being flexible and rich
enough to characterize arbitrary filters in continuous time lag. Moreover, our
method automatically learns the temporal span of the filter. For the particular
application in neuroscience, we designed priors for stimulus and history
filters useful for the spike trains. We compare and validate our method on
simulated and real neural spike train data.
| stat |
AdaFlow: Domain-Adaptive Density Estimator with Application to Anomaly
Detection and Unpaired Cross-Domain Translation | We tackle unsupervised anomaly detection (UAD), a problem of detecting data
that significantly differ from normal data. UAD is typically solved by using
density estimation. Recently, deep neural network (DNN)-based density
estimators, such as Normalizing Flows, have been attracting attention. However,
one of their drawbacks is the difficulty in adapting them to the change in the
normal data's distribution. To address this difficulty, we propose AdaFlow, a
new DNN-based density estimator that can be easily adapted to the change of the
distribution. AdaFlow is a unified model of a Normalizing Flow and Adaptive
Batch-Normalizations, a module that enables DNNs to adapt to new distributions.
AdaFlow can be adapted to a new distribution by just conducting forward
propagation once per sample; hence, it can be used on devices that have limited
computational resources. We have confirmed the effectiveness of the proposed
model through an anomaly detection in a sound task. We also propose a method of
applying AdaFlow to the unpaired cross-domain translation problem, in which one
has to train a cross-domain translation model with only unpaired samples. We
have confirmed that our model can be used for the cross-domain translation
problem through experiments on image datasets.
| stat |
Multiclass Classification via Class-Weighted Nearest Neighbors | We study statistical properties of the k-nearest neighbors algorithm for
multiclass classification, with a focus on settings where the number of classes
may be large and/or classes may be highly imbalanced. In particular, we
consider a variant of the k-nearest neighbor classifier with non-uniform
class-weightings, for which we derive upper and minimax lower bounds on
accuracy, class-weighted risk, and uniform error. Additionally, we show that
uniform error bounds lead to bounds on the difference between empirical
confusion matrix quantities and their population counterparts across a set of
weights. As a result, we may adjust the class weights to optimize
classification metrics such as F1 score or Matthew's Correlation Coefficient
that are commonly used in practice, particularly in settings with imbalanced
classes. We additionally provide a simple example to instantiate our bounds and
numerical experiments.
| stat |
Causal Feature Selection via Orthogonal Search | The problem of inferring the direct causal parents of a response variable
among a large set of explanatory variables is of high practical importance in
many disciplines. Recent work in the field of causal discovery exploits
invariance properties of models across different experimental conditions for
detecting direct causal links. However, these approaches generally do not scale
well with the number of explanatory variables, are difficult to extend to
nonlinear relationships, and require data across different experiments.
Inspired by {\em Debiased} machine learning methods, we study a
one-vs.-the-rest feature selection approach to discover the direct causal
parent of the response. We propose an algorithm that works for purely
observational data, while also offering theoretical guarantees, including the
case of partially nonlinear relationships. Requiring only one estimation for
each variable, we can apply our approach even to large graphs, demonstrating
significant improvements compared to established approaches.
| stat |
An Epidemiological Modelling Approach for Covid19 via Data Assimilation | The global pandemic of the 2019-nCov requires the evaluation of policy
interventions to mitigate future social and economic costs of quarantine
measures worldwide. We propose an epidemiological model for forecasting and
policy evaluation which incorporates new data in real-time through variational
data assimilation. We analyze and discuss infection rates in China, the US and
Italy. In particular, we develop a custom compartmental SIR model fit to
variables related to the epidemic in Chinese cities, named SITR model. We
compare and discuss model results which conducts updates as new observations
become available. A hybrid data assimilation approach is applied to make
results robust to initial conditions. We use the model to do inference on
infection numbers as well as parameters such as the disease transmissibility
rate or the rate of recovery. The parameterisation of the model is parsimonious
and extendable, allowing for the incorporation of additional data and
parameters of interest. This allows for scalability and the extension of the
model to other locations or the adaption of novel data sources.
| stat |
On a Metropolis-Hastings importance sampling estimator | A classical approach for approximating expectations of functions w.r.t.
partially known distributions is to compute the average of function values
along a trajectory of a Metropolis-Hastings (MH) Markov chain. A key part in
the MH algorithm is a suitable acceptance/rejection of a proposed state, which
ensures the correct stationary distribution of the resulting Markov chain.
However, the rejection of proposals causes highly correlated samples. In
particular, when a state is rejected it is not taken any further into account.
In contrast to that we consider a MH importance sampling estimator which
explicitly incorporates all proposed states generated by the MH algorithm. The
estimator satisfies a strong law of large numbers as well as a central limit
theorem, and, in addition to that, we provide an explicit mean squared error
bound. Remarkably, the asymptotic variance of the MH importance sampling
estimator does not involve any correlation term in contrast to its classical
counterpart. Moreover, although the analyzed estimator uses the same amount of
information as the classical MH estimator, it can outperform the latter in
scenarios of moderate dimensions as indicated by numerical experiments.
| stat |
Adjusted Win Ratio with Stratification: Calculation Methods and
Interpretation | The win ratio is a general method of comparing locations of distributions of
two independent, ordinal random variables, and it can be estimated without
distributional assumptions. In this paper we provide a unified theory of win
ratio estimation in the presence of stratification and adjustment by a numeric
variable. Building step by step on the estimate of the crude win ratio we
compare corresponding tests with well known nonparametric tests of group
difference (Wilcoxon rank-sum test, Fligner-Plicello test,
Cochran-Mantel-Haenszel test, test based on the regression on ranks and the
rank ANCOVA test). We show that the win ratio gives an interpretable treatment
effect measure with corresponding test to detect treatment effect difference
under minimal assumptions.
| stat |
A highly likely clusterable data model with no clusters | We propose a model for a dataset in ${\mathbb R}^D$ that does not contain any
clusters but yet is such that a projection of the points on a random
one-dimensional subspace is likely to yield a clustering of the points. This
model is compatible with some recent empirical observations.
| stat |
A Tutorial on Statistically Sound Pattern Discovery | Statistically sound pattern discovery harnesses the rigour of statistical
hypothesis testing to overcome many of the issues that have hampered standard
data mining approaches to pattern discovery. Most importantly, application of
appropriate statistical tests allows precise control over the risk of false
discoveries -- patterns that are found in the sample data but do not hold in
the wider population from which the sample was drawn. Statistical tests can
also be applied to filter out patterns that are unlikely to be useful, removing
uninformative variations of the key patterns in the data. This tutorial
introduces the key statistical and data mining theory and techniques that
underpin this fast developing field.
We concentrate on two general classes of patterns: dependency rules that
express statistical dependencies between condition and consequent parts and
dependency sets that express mutual dependence between set elements. We clarify
alternative interpretations of statistical dependence and introduce appropriate
tests for evaluating statistical significance of patterns in different
situations. We also introduce special techniques for controlling the likelihood
of spurious discoveries when multitudes of patterns are evaluated.
The paper is aimed at a wide variety of audiences. It provides the necessary
statistical background and summary of the state-of-the-art for any data mining
researcher or practitioner wishing to enter or understand statistically sound
pattern discovery research or practice. It can serve as a general introduction
to the field of statistically sound pattern discovery for any reader with a
general background in data sciences.
| stat |
Scalable model selection for spatial additive mixed modeling:
application to crime analysis | A rapid growth in spatial open datasets has led to a huge demand for
regression approaches accommodating spatial and non-spatial effects in big
data. Regression model selection is particularly important to stably estimate
flexible regression models. However, conventional methods can be slow for large
samples. Hence, we develop a fast and practical model-selection approach for
spatial regression models, focusing on the selection of coefficient types that
include constant, spatially varying, and non-spatially varying coefficients. A
pre-processing approach, which replaces data matrices with small inner products
through dimension reduction dramatically accelerates the computation speed of
model selection. Numerical experiments show that our approach selects the model
accurately and computationally efficiently, highlighting the importance of
model selection in the spatial regression context. Then, the present approach
is applied to open data to investigate local factors affecting crime in Japan.
The results suggest that our approach is useful not only for selecting factors
influencing crime risk but also for predicting crime events. This scalable
model selection will be key to appropriately specifying flexible and
large-scale spatial regression models in the era of big data. The developed
model selection approach was implemented in the R package spmoran.
| stat |
Clustering via Ant Colonies: Parameter Analysis and Improvement of the
Algorithm | An ant colony optimization approach for partitioning a set of objects is
proposed. In order to minimize the intra-variance, or within sum-of-squares, of
the partitioned classes, we construct ant-like solutions by a constructive
approach that selects objects to be put in a class with a probability that
depends on the distance between the object and the centroid of the class
(visibility) and the pheromone trail; the latter depends on the class
memberships that have been defined along the iterations. The procedure is
improved with the application of K-means algorithm in some iterations of the
ant colony method. We performed a simulation study in order to evaluate the
method with a Monte Carlo experiment that controls some sensitive parameters of
the clustering problem. After some tuning of the parameters, the method has
also been applied to some benchmark real-data sets. Encouraging results were
obtained in nearly all cases.
| stat |
On the Consistency of Optimal Bayesian Feature Selection in the Presence
of Correlations | Optimal Bayesian feature selection (OBFS) is a multivariate supervised
screening method designed from the ground up for biomarker discovery. In this
work, we prove that Gaussian OBFS is strongly consistent under mild conditions,
and provide rates of convergence for key posteriors in the framework. These
results are of enormous importance, since they identify precisely what features
are selected by OBFS asymptotically, characterize the relative rates of
convergence for posteriors on different types of features, provide conditions
that guarantee convergence, justify the use of OBFS when its internal
assumptions are invalid, and set the stage for understanding the asymptotic
behavior of other algorithms based on the OBFS framework.
| stat |
Hypothesis tests for structured rank correlation matrices | Joint modeling of a large number of variables often requires dimension
reduction strategies that lead to structural assumptions of the underlying
correlation matrix, such as equal pair-wise correlations within subsets of
variables. The underlying correlation matrix is thus of interest for both model
specification and model validation. In this paper, we develop tests of the
hypothesis that the entries of the Kendall rank correlation matrix are linear
combinations of a smaller number of parameters. The asymptotic behaviour of the
proposed test statistics is investigated both when the dimension is fixed and
when it grows with the sample size. We pay special attention to the restricted
hypothesis of partial exchangeability, which contains full exchangeability as a
special case. We show that under partial exchangeability, the test statistics
and their large-sample distributions simplify, which leads to computational
advantages and better performance of the tests. We propose various scalable
numerical strategies for implementation of the proposed procedures, investigate
their finite sample behaviour through simulations, and demonstrate their use on
a real dataset of mean sea levels at various geographical locations.
| stat |
Time delay estimation in satellite imagery time series of precipitation
and NDVI: Pearson's cross correlation revisited | In order to describe more accurately the time relationships between daily
satellite imagery time series of precipitation and NDVI we propose an estimator
which takes into account the sparsity naturally observed in precipitation. We
conducted a series of simulation studies and show that the proposed estimator's
variance is smaller than the canonical's (Pearson-based), in particular, when
the signal-to-noise ratio is rather low. Also, the proposed estimator's
variance was found smaller than the canonical's one when we applied them to
stacks of images (2002-2016) taken on some ecological regions of Mexico.
Computations for this paper are based on functions implemented in our new R
package geoTS.
| stat |
Data-adaptive Dimension Reduction for US Mortality Forecasting | Forecasting accuracy of mortality data is important for the management of
pension funds and pricing of life insurance in actuarial science. Age-specific
mortality forecasting in the US poses a challenging problem in high dimensional
time series analysis. Prior attempts utilize traditional dimension reduction
techniques to avoid the curse of dimensionality, and then mortality forecasting
is achieved through features' forecasting. However, a method of reducing
dimension pertinent to ideal forecasting is elusive. To address this, we
propose a novel approach to pursue features that are not only capable of
representing original data well but also capturing time-serial dependence as
most as possible. The proposed method is adaptive for the US mortality data and
enjoys good statistical performance. As a comparison, our method performs
better than existing approaches, especially in regard to the Lee-Carter Model
as a benchmark in mortality analysis. Based on forecasting results, we generate
more accurate estimates of future life expectancies and prices of life
annuities, which can have great financial impact on life insurers and social
securities compared with using Lee-Carter Model. Furthermore, various
simulations illustrate scenarios under which our method has advantages, as well
as interpretation of the good performance on mortality data.
| stat |
A flexible EM-like clustering algorithm for noisy data | Though very popular, it is well known that the EM for GMM algorithm suffers
from non-Gaussian distribution shapes, outliers and high-dimensionality. In
this paper, we design a new robust clustering algorithm that can efficiently
deal with noise and outliers in diverse data sets. As an EM-like algorithm, it
is based on both estimations of clusters centers and covariances. In addition,
using a semi-parametric paradigm, the method estimates an unknown scale
parameter per data-point. This allows the algorithm to accommodate for heavier
tails distributions and outliers without significantly loosing efficiency in
various classical scenarios. We first derive and analyze the proposed algorithm
in the context of elliptical distributions, showing in particular important
insensitivity properties to the underlying data distributions. We then study
the convergence and accuracy of the algorithm by considering first synthetic
data. Then, we show that the proposed algorithm outperforms other classical
unsupervised methods of the literature such as k-means, the EM for Gaussian
mixture models and its recent modifications or spectral clustering when applied
to real data sets as MNIST, NORB, and 20newsgroups.
| stat |
A statistical framework for measuring the temporal stability of human
mobility patterns | Despite the growing popularity of human mobility studies that collect GPS
location data, the problem of determining the minimum required length of GPS
monitoring has not been addressed in the current statistical literature. In
this paper we tackle this problem by laying out a theoretical framework for
assessing the temporal stability of human mobility based on GPS location data.
We define several measures of the temporal dynamics of human spatiotemporal
trajectories based on the average velocity process, and on activity
distributions in a spatial observation window. We demonstrate the use of our
methods with data that comprise the GPS locations of 185 individuals over the
course of 18 months. Our empirical results suggest that GPS monitoring should
be performed over periods of time that are significantly longer than what has
been previously suggested. Furthermore, we argue that GPS study designs should
take into account demographic groups.
KEYWORDS: Density estimation; global positioning systems (GPS); human
mobility; spatiotemporal trajectories; temporal dynamics
| stat |
A Mixture Model to Detect Edges in Sparse Co-expression Graphs | In the early days of gene expression data, researchers have focused on
gene-level analysis, and particularly on finding differentially expressed
genes. This usually involved making a simplifying assumption that genes are
independent, which made likelihood derivations feasible and allowed for
relatively simple implementations. In recent years, the scope has expanded to
include pathway and `gene set' analysis in an attempt to understand the
relationships between genes. We develop a method to recover a gene network's
structure from co-expression data, measured in terms of normalized Pearson's
correlation coefficients between gene pairs. We treat these co-expression
measurements as weights in the complete graph in which nodes correspond to
genes. To decide which edges exist in the gene network, we fit a
three-component mixture model such that the observed weights of `null edges'
follow a normal distribution with mean 0, and the non-null edges follow a
mixture of two lognormal distributions, one for positively- and one for
negatively-correlated pairs. We show that this so-called L2N mixture model
outperforms other methods in terms of power to detect edges, and it allows us
to control the false discovery rate. Importantly, the method makes no
assumptions about the true network structure.
| stat |
A relation between log-likelihood and cross-validation log-scores | It is shown that the log-likelihood of a hypothesis or model given some data
is equivalent to an average of all leave-one-out cross-validation log-scores
that can be calculated from all subsets of the data. This relation can be
generalized to any $k$-fold cross-validation log-scores.
| stat |
A Frequency Domain Bootstrap for General Multivariate Stationary
Processes | For many relevant statistics of multivariate time series, no valid frequency
domain bootstrap procedures exist. This is mainly due to the fact that the
distribution of such statistics depends on the fourth-order moment structure of
the underlying process in nearly every scenario, except for some special cases
like Gaussian time series. In contrast to the univariate case, even additional
structural assumptions such as linearity of the multivariate process or a
standardization of the statistic of interest do not solve the problem. This
paper focuses on integrated periodogram statistics as well as functions thereof
and presents a new frequency domain bootstrap procedure for multivariate time
series, the multivariate frequency domain hybrid bootstrap (MFHB), to fill this
gap. Asymptotic validity of the MFHB procedure is established for general
classes of periodogram-based statistics and for stationary multivariate
processes satisfying rather weak dependence conditions. A simulation study is
carried out which compares the finite sample performance of the MFHB with that
of the moving block bootstrap.
| stat |
Multilevel Delayed Acceptance MCMC with an Adaptive Error Model in PyMC3 | Uncertainty Quantification through Markov Chain Monte Carlo (MCMC) can be
prohibitively expensive for target probability densities with expensive
likelihood functions, for instance when the evaluation it involves solving a
Partial Differential Equation (PDE), as is the case in a wide range of
engineering applications. Multilevel Delayed Acceptance (MLDA) with an Adaptive
Error Model (AEM) is a novel approach, which alleviates this problem by
exploiting a hierarchy of models, with increasing complexity and cost, and
correcting the inexpensive models on-the-fly. The method has been integrated
within the open-source probabilistic programming package PyMC3 and is available
in the latest development version. In this paper, the algorithm is presented
along with an illustrative example.
| stat |
Neural Bayes: A Generic Parameterization Method for Unsupervised
Representation Learning | We introduce a parameterization method called Neural Bayes which allows
computing statistical quantities that are in general difficult to compute and
opens avenues for formulating new objectives for unsupervised representation
learning. Specifically, given an observed random variable $\mathbf{x}$ and a
latent discrete variable $z$, we can express $p(\mathbf{x}|z)$,
$p(z|\mathbf{x})$ and $p(z)$ in closed form in terms of a sufficiently
expressive function (Eg. neural network) using our parameterization without
restricting the class of these distributions. To demonstrate its usefulness, we
develop two independent use cases for this parameterization:
1. Mutual Information Maximization (MIM): MIM has become a popular means for
self-supervised representation learning. Neural Bayes allows us to compute
mutual information between observed random variables $\mathbf{x}$ and latent
discrete random variables $z$ in closed form. We use this for learning image
representations and show its usefulness on downstream classification tasks.
2. Disjoint Manifold Labeling: Neural Bayes allows us to formulate an
objective which can optimally label samples from disjoint manifolds present in
the support of a continuous distribution. This can be seen as a specific form
of clustering where each disjoint manifold in the support is a separate
cluster. We design clustering tasks that obey this formulation and empirically
show that the model optimally labels the disjoint manifolds. Our code is
available at \url{https://github.com/salesforce/NeuralBayes}
| stat |
Identification of Linear Regressions with Errors in all Variables | This paper analyzes the classical linear regression model with measurement
errors in all the variables. First, we provide necessary and sufficient
conditions for identification of the coefficients. We show that the
coefficients are not identified if and only if an independent normally
distributed linear combination of regressors can be transferred from the
regressors to the errors. Second, we introduce a new estimator for the
coefficients using a continuum of moments that are based on second derivatives
of the log characteristic function of the observables. In Monte Carlo
simulations, the estimator performs well and is robust to the amount of
measurement error and number of mismeasured regressors. In an application to
firm investment decisions, the estimates are similar to those produced by a
generalized method of moments estimator based on third to fifth moments.
| stat |
The decoupled extended Kalman filter for dynamic exponential-family
factorization models | Motivated by the needs of online large-scale recommender systems, we
specialize the decoupled extended Kalman filter (DEKF) to factorization models,
including factorization machines, matrix and tensor factorization, and
illustrate the effectiveness of the approach through numerical experiments on
synthetic and on real-world data. Online learning of model parameters through
the DEKF makes factorization models more broadly useful by (i) allowing for
more flexible observations through the entire exponential family, (ii) modeling
parameter drift, and (iii) producing parameter uncertainty estimates that can
enable explore/exploit and other applications. We use a different parameter
dynamics than the standard DEKF, allowing parameter drift while encouraging
reasonable values. We also present an alternate derivation of the extended
Kalman filter and DEKF that highlights the role of the Fisher information
matrix in the EKF.
| stat |
Causal inference and machine learning approaches for evaluation of the
health impacts of large-scale air quality regulations | We develop a causal inference approach to estimate the number of adverse
health events prevented by large-scale air quality regulations via changes in
exposure to multiple pollutants. This approach is motivated by regulations that
impact pollution levels in all areas within their purview. We introduce a
causal estimand called the Total Events Avoided (TEA) by the regulation,
defined as the difference in the expected number of health events under the
no-regulation pollution exposures and the observed number of health events
under the with-regulation pollution exposures. We propose a matching method and
a machine learning method that leverage high-resolution, population-level
pollution and health data to estimate the TEA. Our approach improves upon
traditional methods for regulation health impact analyses by clarifying the
causal identifying assumptions, utilizing population-level data, minimizing
parametric assumptions, and considering the impacts of multiple pollutants
simultaneously. To reduce model-dependence, the TEA estimate captures health
impacts only for units in the data whose anticipated no-regulation features are
within the support of the observed with-regulation data, thereby providing a
conservative but data-driven assessment to complement traditional parametric
approaches. We apply these methods to investigate the health impacts of the
1990 Clean Air Act Amendments in the US Medicare population.
| stat |
The Ockham's razor applied to COVID-19 model fitting French data | This paper presents a data-based simple model for fitting the available data
of the Covid-19 pandemic evolution in France. The time series concerning the 13
regions of mainland France have been considered for fitting and validating the
model. An extremely simple, two-dimensional model with only two parameters
demonstrated to be able to reproduce the time series concerning the number of
daily demises caused by Covid-19, the hospitalizations, intensive care and
emergency accesses, the daily number of positive test and other indicators, for
the different French regions. These results might contribute to stimulate a
debate on the suitability of much more complex models for reproducing and
forecasting the pandemic evolution since, although relevant from a mechanistic
point of view, they could lead to nonidentifiability issues.
| stat |
Spread Divergences | For distributions p and q with different supports, the divergence D(p|q) may
not exist. We define a spread divergence on modified p and q and describe
sufficient conditions for the existence of such a divergence. We demonstrate
how to maximize the discriminatory power of a given divergence by
parameterizing and learning the spread. We also give examples of using a spread
divergence to train and improve implicit generative models, including linear
models (Independent Components Analysis) and non-linear models (Deep Generative
Networks).
| stat |
A Deep Learning Approach for Predicting Spatiotemporal Dynamics From
Sparsely Observed Data | In this paper, we consider the problem of learning prediction models for
spatiotemporal physical processes driven by unknown partial differential
equations (PDEs). We propose a deep learning framework that learns the
underlying dynamics and predicts its evolution using sparsely distributed data
sites. Deep learning has shown promising results in modeling physical dynamics
in recent years. However, most of the existing deep learning methods for
modeling physical dynamics either focus on solving known PDEs or require data
in a dense grid when the governing PDEs are unknown. In contrast, our method
focuses on learning prediction models for unknown PDE-driven dynamics only from
sparsely observed data. The proposed method is spatial dimension-independent
and geometrically flexible. We demonstrate our method in the forecasting task
for the two-dimensional wave equation and the Burgers-Fisher equation in
multiple geometries with different boundary conditions, and the ten-dimensional
heat equation.
| stat |
Significance testing for canonical correlation analysis in high
dimensions | We consider the problem of testing for the presence of linear relationships
between large sets of random variables based on a post-selection inference
approach to canonical correlation analysis. The challenge is to adjust for the
selection of subsets of variables having linear combinations with maximal
sample correlation. To this end, we construct a stabilized one-step estimator
of the euclidean-norm of the canonical correlations maximized over subsets of
variables of pre-specified cardinality. This estimator is shown to be
consistent for its target parameter and asymptotically normal provided the
dimensions of the variables do not grow too quickly with sample size. We also
develop a greedy search algorithm to accurately compute the estimator, leading
to a computationally tractable omnibus test for the global null hypothesis that
there are no linear relationships between any subsets of variables having the
pre-specified cardinality. Further, we develop a confidence interval for the
target parameter that takes the variable selection into account.
| stat |
Propensity Score Estimation Using Density Ratio Model under Item
Nonresponse | Missing data is frequently encountered in practice. Propensity score
estimation is a popular tool for handling such missingness. The propensity
score is often developed using the model for the response probability, which
can be subject to model misspecification. In this paper, we consider an
alternative approach of estimating the inverse of the propensity scores using
the density ratio function. By partitioning the sample into two groups based on
the response status of the elements, we can apply the density ratio function
estimation method and obtain the inverse propensity scores for nonresponse
adjustment. Density ratio estimation can be obtained by applying the so-called
maximum entropy method, which uses the Kullback-Leibler divergence measure
under calibration constraints. By including the covariates for the outcome
regression models only into the density ratio model, we can achieve efficient
propensity score estimation. We further extend the proposed approach to the
multivariate missing case. Some limited simulation studies are presented to
compare with the existing methods.
| stat |
On Uniform Convergence and Low-Norm Interpolation Learning | We consider an underdetermined noisy linear regression model where the
minimum-norm interpolating predictor is known to be consistent, and ask: can
uniform convergence in a norm ball, or at least (following Nagarajan and
Kolter) the subset of a norm ball that the algorithm selects on a typical input
set, explain this success? We show that uniformly bounding the difference
between empirical and population errors cannot show any learning in the norm
ball, and cannot show consistency for any set, even one depending on the exact
algorithm and distribution. But we argue we can explain the consistency of the
minimal-norm interpolator with a slightly weaker, yet standard, notion: uniform
convergence of zero-error predictors in a norm ball. We use this to bound the
generalization error of low- (but not minimal-) norm interpolating predictors.
| stat |
Likelihood-based inference for modelling packet transit from thinned
flow summaries | The substantial growth of network traffic speed and volume presents practical
challenges to network data analysis. Packet thinning and flow aggregation
protocols such as NetFlow reduce the size of datasets by providing structured
data summaries, but conversely this impedes statistical inference. Methods
which aim to model patterns of traffic propagation typically do not account for
the packet thinning and summarisation process into the analysis, and are often
simplistic, e.g.~method-of-moments. As a result, they can be of limited
practical use.
We introduce a likelihood-based analysis which fully incorporates packet
thinning and NetFlow summarisation into the analysis. As a result, inferences
can be made for models on the level of individual packets while only observing
thinned flow summary information. We establish consistency of the resulting
maximum likelihood estimator, derive bounds on the volume of traffic which
should be observed to achieve required levels of estimator accuracy, and
identify an ideal family of models. The robust performance of the estimator is
examined through simulated analyses and an application on a publicly available
trace dataset containing over 36m packets over a 1 minute period.
| stat |
Multiple system estimation using covariates having missing values and
measurement error: estimating the size of the M\=aori population in New
Zealand | We investigate use of two or more linked registers, or lists, for both
population size estimation and to investigate the relationship between
variables appearing on all or only some registers. This relationship is usually
not fully known because some individuals appear in only some registers, and
some are not in any register. These two problems have been solved
simultaneously using the EM algorithm. We extend this approach to estimate the
size of the indigenous M\=aori population in New Zealand, leading to several
innovations: (1) the approach is extended to four registers (including the
population census), where the reporting of M\=aori status differs between
registers; (2) some individuals in one or more registers have missing
ethnicity, and we adapt the approach to handle this additional missingness; (3)
some registers cover subsets of the population by design. We discuss under
which assumptions such structural undercoverage can be ignored and provide a
general result; (4) we treat the M\=aori indicator in each register as a
variable measured with error, and embed a latent class model in the multiple
system estimation to estimate the population size of a latent variable,
interpreted as the true M\=aori status. Finally, we discuss estimating the
M\=aori population size from administrative data only. Supplementary materials
for our article are available online.
| stat |
Non-linear process convolutions for multi-output Gaussian processes | The paper introduces a non-linear version of the process convolution
formalism for building covariance functions for multi-output Gaussian
processes. The non-linearity is introduced via Volterra series, one series per
each output. We provide closed-form expressions for the mean function and the
covariance function of the approximated Gaussian process at the output of the
Volterra series. The mean function and covariance function for the joint
Gaussian process are derived using formulae for the product moments of Gaussian
variables. We compare the performance of the non-linear model against the
classical process convolution approach in one synthetic dataset and two real
datasets.
| stat |
Learning Multi-layer Latent Variable Model via Variational Optimization
of Short Run MCMC for Approximate Inference | This paper studies the fundamental problem of learning deep generative models
that consist of multiple layers of latent variables organized in top-down
architectures. Such models have high expressivity and allow for learning
hierarchical representations. Learning such a generative model requires
inferring the latent variables for each training example based on the posterior
distribution of these latent variables. The inference typically requires Markov
chain Monte Caro (MCMC) that can be time consuming. In this paper, we propose
to use noise initialized non-persistent short run MCMC, such as finite step
Langevin dynamics initialized from the prior distribution of the latent
variables, as an approximate inference engine, where the step size of the
Langevin dynamics is variationally optimized by minimizing the Kullback-Leibler
divergence between the distribution produced by the short run MCMC and the
posterior distribution. Our experiments show that the proposed method
outperforms variational auto-encoder (VAE) in terms of reconstruction error and
synthesis quality. The advantage of the proposed method is that it is simple
and automatic without the need to design an inference model.
| stat |
Conformal prediction interval for dynamic time-series | We develop a method to construct distribution-free prediction intervals for
dynamic time-series, called \Verb|EnbPI| that wraps around any bootstrap
ensemble estimator to construct sequential prediction intervals. \Verb|EnbPI|
is closely related to the conformal prediction (CP) framework but does not
require data exchangeability. Theoretically, these intervals attain
finite-sample, \textit{approximately valid} marginal coverage for broad classes
of regression functions and time-series with strongly mixing stochastic errors.
Computationally, \Verb|EnbPI| avoids overfitting and requires neither
data-splitting nor training multiple ensemble estimators; it efficiently
aggregates bootstrap estimators that have been trained. In general,
\Verb|EnbPI| is easy to implement, scalable to producing arbitrarily many
prediction intervals sequentially, and well-suited to a wide range of
regression functions. We perform extensive real-data analyses to demonstrate
its effectiveness.
| stat |
Model Averaging Estimation for Partially Linear Functional Score Models | This paper is concerned with model averaging estimation for partially linear
functional score models. These models predict a scalar response using both
parametric effect of scalar predictors and non-parametric effect of a
functional predictor. Within this context, we develop a Mallows-type criterion
for choosing weights. The resulting model averaging estimator is proved to be
asymptotically optimal under certain regularity conditions in terms of
achieving the smallest possible squared error loss. Simulation studies
demonstrate its superiority or comparability to information criterion
score-based model selection and averaging estimators. The proposed procedure is
also applied to two real data sets for illustration. That the components of
nonparametric part are unobservable leads to a more complicated situation than
ordinary partially linear models (PLM) and a different theoretical derivation
from those of PLM.
| stat |
Estimating Spatially-Smoothed Fiber Orientation Distribution | Diffusion-weighted magnetic resonance imaging (D-MRI) is an in-vivo and
non-invasive imaging technology to probe anatomical architectures of biological
samples. The anatomy of white matter fiber tracts in the brain can be revealed
to help understanding of the connectivity patterns among different brain
regions. In this paper, we propose a novel Nearest-neighbor Adaptive Regression
Model (NARM) for adaptive estimation of the fiber orientation distribution
(FOD) function based on D-MRI data, where spatial homogeneity is used to
improve FOD estimation by incorporating neighborhood information. Specifically,
we formulate the FOD estimation problem as a weighted linear regression
problem, where the weights are chosen to account for spatial proximity and
potential heterogeneity due to different fiber configurations. The weights are
adaptively updated and a stopping rule based on nearest neighbor distance is
designed to prevent over-smoothing. NARM is further extended to accommodate
D-MRI data with multiple bvalues.
Comprehensive simulation results demonstrate that NARM leads to satisfactory
FOD reconstructions and performs better than voxel-wise estimation as well as
competing smoothing methods. By applying NARM to real 3T D-MRI datasets, we
demonstrate the effectiveness of NARM in recovering more realistic crossing
fiber patterns and producing more coherent fiber tracking results, establishing
the practical value of NARM for analyzing D-MRI data and providing reliable
information on brain structural connectivity.
| stat |
Variable selection and estimation for the additive hazards model subject
to left-truncation, right-censoring and measurement error in covariates | High-dimensional sparse modeling with censored survival data is of great
practical importance, and several methods have been proposed for variable
selection based on different models. However, the impact of biased sample
caused by left-truncation and covariates measurement error to variable
selection is not fully explored. In this paper, we mainly focus on the additive
hazards model and analyze the high-dimensional survival data subject to
left-truncation and measurement error in covariates. We develop the three-stage
procedure to correct the error effect, select variables, and estimate the
parameters of interest simultaneously. Numerical studies are reported to assess
the performance of the proposed methods.
| stat |
Implicit regularization and solution uniqueness in over-parameterized
matrix sensing | We consider whether algorithmic choices in over-parameterized linear matrix
factorization introduce implicit regularization. We focus on noiseless matrix
sensing over rank-$r$ positive semi-definite (PSD) matrices in $\mathbb{R}^{n
\times n}$, with a sensing mechanism that satisfies restricted isometry
properties (RIP). The algorithm we study is \emph{factored gradient descent},
where we model the low-rankness and PSD constraints with the factorization
$UU^\top$, for $U \in \mathbb{R}^{n \times r}$. Surprisingly, recent work
argues that the choice of $r \leq n$ is not pivotal: even setting $U \in
\mathbb{R}^{n \times n}$ is sufficient for factored gradient descent to find
the rank-$r$ solution, which suggests that operating over the factors leads to
an implicit regularization. In this contribution, we provide a different
perspective to the problem of implicit regularization. We show that under
certain conditions, the PSD constraint by itself is sufficient to lead to a
unique rank-$r$ matrix recovery, without implicit or explicit low-rank
regularization. \emph{I.e.}, under assumptions, the set of PSD matrices, that
are consistent with the observed data, is a singleton, regardless of the
algorithm used.
| stat |
Probabilistic Machine Learning for Healthcare | Machine learning can be used to make sense of healthcare data. Probabilistic
machine learning models help provide a complete picture of observed data in
healthcare. In this review, we examine how probabilistic machine learning can
advance healthcare. We consider challenges in the predictive model building
pipeline where probabilistic models can be beneficial including calibration and
missing data. Beyond predictive models, we also investigate the utility of
probabilistic machine learning models in phenotyping, in generative models for
clinical use cases, and in reinforcement learning.
| stat |
Conditional deep surrogate models for stochastic, high-dimensional, and
multi-fidelity systems | We present a probabilistic deep learning methodology that enables the
construction of predictive data-driven surrogates for stochastic systems.
Leveraging recent advances in variational inference with implicit
distributions, we put forth a statistical inference framework that enables the
end-to-end training of surrogate models on paired input-output observations
that may be stochastic in nature, originate from different information sources
of variable fidelity, or be corrupted by complex noise processes. The resulting
surrogates can accommodate high-dimensional inputs and outputs and are able to
return predictions with quantified uncertainty. The effectiveness our approach
is demonstrated through a series of canonical studies, including the regression
of noisy data, multi-fidelity modeling of stochastic processes, and uncertainty
propagation in high-dimensional dynamical systems.
| stat |
Forecasting high-dimensional dynamics exploiting suboptimal embeddings | Delay embedding---a method for reconstructing dynamical systems by delay
coordinates---is widely used to forecast nonlinear time series as a model-free
approach. When multivariate time series are observed, several existing
frameworks can be applied to yield a single forecast combining multiple
forecasts derived from various embeddings. However, the performance of these
frameworks is not always satisfactory because they randomly select embeddings
or use brute force and do not consider the diversity of the embeddings to
combine. Herein, we develop a forecasting framework that overcomes these
existing problems. The framework exploits various "suboptimal embeddings"
obtained by minimizing the in-sample error via combinatorial optimization. The
framework achieves the best results among existing frameworks for sample toy
datasets and a real-world flood dataset. We show that the framework is
applicable to a wide range of data lengths and dimensions. Therefore, the
framework can be applied to various fields such as neuroscience, ecology,
finance, fluid dynamics, weather, and disaster prevention.
| stat |
Targeted Estimation of L2 Distance Between Densities and its Application
to Geo-spatial Data | We examine the integrated squared difference, also known as the L2 distance
(L2D), between two probability densities. Such a distance metric allows for
comparison of differences between pairs of distributions or changes in a
distribution over time. We propose a targeted maximum likelihood estimator for
this parameter based on samples of independent and identically distributed
observations from both underlying distributions. We compare our method to
kernel density estimation and demonstrate superior performance for our method
with regards to confidence interval coverage rate and mean squared error.
| stat |
A Bayesian Fisher-EM algorithm for discriminative Gaussian subspace
clustering | High-dimensional data clustering has become and remains a challenging task
for modern statistics and machine learning, with a wide range of applications.
We consider in this work the powerful discriminative latent mixture model, and
we extend it to the Bayesian framework. Modeling data as a mixture of Gaussians
in a low-dimensional discriminative subspace, a Gaussian prior distribution is
introduced over the latent group means and a family of twelve submodels are
derived considering different covariance structures. Model inference is done
with a variational EM algorithm, while the discriminative subspace is estimated
via a Fisher-step maximizing an unsupervised Fisher criterion. An empirical
Bayes procedure is proposed for the estimation of the prior hyper-parameters,
and an integrated classification likelihood criterion is derived for selecting
both the number of clusters and the submodel. The performances of the resulting
Bayesian Fisher-EM algorithm are investigated in two thorough simulated
scenarios, regarding both dimensionality as well as noise and assessing its
superiority with respect to state-of-the-art Gaussian subspace clustering
models. In addition to standard real data benchmarks, an application to single
image denoising is proposed, displaying relevant results. This work comes with
a reference implementation for the R software in the FisherEM package
accompanying the paper.
| stat |
Approximate spectral gaps for Markov chains mixing times in high
dimensions | This paper introduces a concept of approximate spectral gap to analyze the
mixing time of Markov Chain Monte Carlo (MCMC) algorithms for which the usual
spectral gap is degenerate or almost degenerate. We use the idea to analyze a
class of MCMC algorithms to sample from mixtures of densities. As an
application we study the mixing time of a Gibbs sampler for variable selection
in linear regression models. Under some regularity conditions on the signal and
the design matrix of the regression problem, we show that for well-chosen
initial distributions the mixing time of the Gibbs sampler is polynomial in the
dimension of the space.
| stat |
How to capitalize on a priori contrasts in linear (mixed) models: A
tutorial | Factorial experiments in research on memory, language, and in other areas are
often analyzed using analysis of variance (ANOVA). However, for effects with
more than one numerator degrees of freedom, e.g., for experimental factors with
more than two levels, the ANOVA omnibus F-test is not informative about the
source of a main effect or interaction. Because researchers typically have
specific hypotheses about which condition means differ from each other, a
priori contrasts (i.e., comparisons planned before the sample means are known)
between specific conditions or combinations of conditions are the appropriate
way to represent such hypotheses in the statistical model. Many researchers
have pointed out that contrasts should be "tested instead of, rather than as a
supplement to, the ordinary `omnibus' F test" (Hays, 1973, p. 601). In this
tutorial, we explain the mathematics underlying different kinds of contrasts
(i.e., treatment, sum, repeated, polynomial, custom, nested, interaction
contrasts), discuss their properties, and demonstrate how they are applied in
the R System for Statistical Computing (R Core Team, 2018). In this context, we
explain the generalized inverse which is needed to compute the coefficients for
contrasts that test hypotheses that are not covered by the default set of
contrasts. A detailed understanding of contrast coding is crucial for
successful and correct specification in linear models (including linear mixed
models). Contrasts defined a priori yield far more useful confirmatory tests of
experimental hypotheses than standard omnibus F-test. Reproducible code is
available from https://osf.io/7ukf6/.
| stat |
Calibrate, Emulate, Sample | Many parameter estimation problems arising in applications are best cast in
the framework of Bayesian inversion. This allows not only for an estimate of
the parameters, but also for the quantification of uncertainties in the
estimates. Often in such problems the parameter-to-data map is very expensive
to evaluate, and computing derivatives of the map, or derivative-adjoints, may
not be feasible. Additionally, in many applications only noisy evaluations of
the map may be available. We propose an approach to Bayesian inversion in such
settings that builds on the derivative-free optimization capabilities of
ensemble Kalman inversion methods. The overarching approach is to first use
ensemble Kalman sampling (EKS) to calibrate the unknown parameters to fit the
data; second, to use the output of the EKS to emulate the parameter-to-data
map; third, to sample from an approximate Bayesian posterior distribution in
which the parameter-to-data map is replaced by its emulator. This results in a
principled approach to approximate Bayesian inference that requires only a
small number of evaluations of the (possibly noisy approximation of the)
parameter-to-data map. It does not require derivatives of this map, but instead
leverages the documented power of ensemble Kalman methods. Furthermore, the EKS
has the desirable property that it evolves the parameter ensembles towards the
regions in which the bulk of the parameter posterior mass is located, thereby
locating them well for the emulation phase of the methodology. In essence, the
EKS methodology provides a cheap solution to the design problem of where to
place points in parameter space to efficiently train an emulator of the
parameter-to-data map for the purposes of Bayesian inversion.
| stat |
Reconstructing regime-dependent causal relationships from observational
time series | Inferring causal relations from observational time series data is a key
problem across science and engineering whenever experimental interventions are
infeasible or unethical. Increasing data availability over the past decades has
spurred the development of a plethora of causal discovery methods, each
addressing particular challenges of this difficult task. In this paper we focus
on an important challenge that is at the core of time series causal discovery:
regime-dependent causal relations. Often dynamical systems feature transitions
depending on some, often persistent, unobserved background regime, and
different regimes may exhibit different causal relations. Here, we assume a
persistent and discrete regime variable leading to a finite number of regimes
within which we may assume stationary causal relations. To detect
regime-dependent causal relations, we combine the conditional
independence-based PCMCI method with a regime learning optimisation approach.
PCMCI allows for linear and nonlinear, high-dimensional time series causal
discovery. Our method, Regime-PCMCI, is evaluated on a number of numerical
experiments demonstrating that it can distinguish regimes with different causal
directions, time lags, effects and sign of causal links, as well as changes in
the variables' autocorrelation. Further, Regime-PCMCI is employed to
observations of El Ni\~no Southern Oscillation and Indian rainfall,
demonstrating skill also in real-world datasets.
| stat |
A smooth dynamic network model for patent collaboration data | The development and application of models, which take the evolution of
network dynamics into account are receiving increasing attention. We contribute
to this field and focus on a profile likelihood approach to model time-stamped
event data for a large-scale dynamic network. We investigate the collaboration
of inventors using EU patent data. As event we consider the submission of a
joint patent and we explore the driving forces for collaboration between
inventors. We propose a flexible semiparametric model, which includes external
and internal covariates, where the latter are built from the network history.
| stat |
Multivariate spectral downscaling for PM2.5 species | Fine particulate matter (PM2.5) is a mixture of air pollutants that has
adverse effects on human health. Understanding the health effects of PM2.5
mixture and its individual species has been a research priority over the past
two decades. However, the limited availability of speciated PM2.5 measurements
continues to be a major challenge in exposure assessment for conducting
large-scale population-based epidemiology studies. The PM2.5 species have
complex spatial-temporal and cross dependence structures that should be
accounted for in estimating the spatiotemporal distribution of each component.
Two major sources of air quality data are commonly used for deriving exposure
estimates: point-level monitoring data and gridded numerical computer model
simulation, such as the Community Multiscale Air Quality (CMAQ) model. We
propose a statistical method to combine these two data sources for estimating
speciated PM2.5 concentration. Our method models the complex relationships
between monitoring measurements and the numerical model output at different
spatial resolutions, and we model the spatial dependence and cross dependence
among PM2.5 species. We apply the method to combine CMAQ model output with
major PM2.5 species measurements in the contiguous United States in 2011.
| stat |
Estimation and Inference for High Dimensional Generalized Linear Models:
A Splitting and Smoothing Approach | The focus of modern biomedical studies has gradually shifted to explanation
and estimation of joint effects of high dimensional predictors on disease
risks. Quantifying uncertainty in these estimates may provide valuable insight
into prevention strategies or treatment decisions for both patients and
physicians. High dimensional inference, including confidence intervals and
hypothesis testing, has sparked much interest. While much work has been done in
the linear regression setting, there is lack of literature on inference for
high dimensional generalized linear models. We propose a novel and
computationally feasible method, which accommodates a variety of outcome types,
including normal, binomial, and Poisson data. We use a "splitting and
smoothing" approach, which splits samples into two parts, performs variable
selection using one part and conducts partial regression with the other part.
Averaging the estimates over multiple random splits, we obtain the smoothed
estimates, which are numerically stable. We show that the estimates are
consistent, asymptotically normal, and construct confidence intervals with
proper coverage probabilities for all predictors. We examine the finite sample
performance of our method by comparing it with the existing methods and
applying it to analyze a lung cancer cohort study.
| stat |
Contrastive latent variable modeling with application to case-control
sequencing experiments | High-throughput RNA-sequencing (RNA-seq) technologies are powerful tools for
understanding cellular state. Often it is of interest to quantify and summarize
changes in cell state that occur between experimental or biological conditions.
Differential expression is typically assessed using univariate tests to measure
gene-wise shifts in expression. However, these methods largely ignore changes
in transcriptional correlation. Furthermore, there is a need to identify the
low-dimensional structure of the gene expression shift to identify collections
of genes that change between conditions. Here, we propose contrastive latent
variable models designed for count data to create a richer portrait of
differential expression in sequencing data. These models disentangle the
sources of transcriptional variation in different conditions, in the context of
an explicit model of variation at baseline. Moreover, we develop a model-based
hypothesis testing framework that can test for global and gene subset-specific
changes in expression. We test our model through extensive simulations and
analyses with count-based gene expression data from perturbation and
observational sequencing experiments. We find that our methods can effectively
summarize and quantify complex transcriptional changes in case-control
experimental sequencing data.
| stat |
Norm matters: efficient and accurate normalization schemes in deep
networks | Over the past few years, Batch-Normalization has been commonly used in deep
networks, allowing faster training and high performance for a wide variety of
applications. However, the reasons behind its merits remained unanswered, with
several shortcomings that hindered its use for certain tasks. In this work, we
present a novel view on the purpose and function of normalization methods and
weight-decay, as tools to decouple weights' norm from the underlying optimized
objective. This property highlights the connection between practices such as
normalization, weight decay and learning-rate adjustments. We suggest several
alternatives to the widely used $L^2$ batch-norm, using normalization in $L^1$
and $L^\infty$ spaces that can substantially improve numerical stability in
low-precision implementations as well as provide computational and memory
benefits. We demonstrate that such methods enable the first batch-norm
alternative to work for half-precision implementations. Finally, we suggest a
modification to weight-normalization, which improves its performance on
large-scale tasks.
| stat |
Cost-sensitive Selection of Variables by Ensemble of Model Sequences | Many applications require the collection of data on different variables or
measurements over many system performance metrics. We term those broadly as
measures or variables. Often data collection along each measure incurs a cost,
thus it is desirable to consider the cost of measures in modeling. This is a
fairly new class of problems in the area of cost-sensitive learning. A few
attempts have been made to incorporate costs in combining and selecting
measures. However, existing studies either do not strictly enforce a budget
constraint, or are not the `most' cost effective. With a focus on
classification problem, we propose a computationally efficient approach that
could find a near optimal model under a given budget by exploring the most
`promising' part of the solution space. Instead of outputting a single model,
we produce a model schedule -- a list of models, sorted by model costs and
expected predictive accuracy. This could be used to choose the model with the
best predictive accuracy under a given budget, or to trade off between the
budget and the predictive accuracy. Experiments on some benchmark datasets show
that our approach compares favorably to competing methods.
| stat |
BAMBI: An R package for Fitting Bivariate Angular Mixture Models | Statistical analyses of directional or angular data have applications in a
variety of fields, such as geology, meteorology and bioinformatics. There is
substantial literature on descriptive and inferential techniques for univariate
angular data, with the bivariate (or more generally, multivariate) cases
receiving more attention in recent years. More specifically, the bivariate
wrapped normal, von Mises sine and von Mises cosine distributions, and mixtures
thereof, have been proposed for practical use. However, there is a lack of
software implementing these distributions and the associated inferential
techniques. In this article, we introduce BAMBI, an R package for analyzing
bivariate (and univariate) angular data. We implement random data generation,
density evaluation, and computation of theoretical summary measures (variances
and correlation coefficients) for the three aforementioned bivariate angular
distributions, as well as two univariate angular distributions: the univariate
wrapped normal and the univariate von Mises distribution. The major
contribution of BAMBI to statistical computing is in providing Bayesian methods
for modeling angular data using finite mixtures of these distributions. We also
provide functions for visual and numerical diagnostics and Bayesian inference
for the fitted models. In this article, we first provide a brief review of the
distributions and techniques used in BAMBI, then describe the capabilities of
the package, and finally conclude with demonstrations of mixture model fitting
using BAMBI on the two real datasets included in the package, one univariate
and one bivariate.
| stat |
Finding the bandit in a graph: Sequential search-and-stop | We consider the problem where an agent wants to find a hidden object that is
randomly located in some vertex of a directed acyclic graph (DAG) according to
a fixed but possibly unknown distribution. The agent can only examine vertices
whose in-neighbors have already been examined. In this paper, we address a
learning setting where we allow the agent to stop before having found the
object and restart searching on a new independent instance of the same problem.
Our goal is to maximize the total number of hidden objects found given a time
budget. The agent can thus skip an instance after realizing that it would spend
too much time on it. Our contributions are both to the search theory and
multi-armed bandits. If the distribution is known, we provide a quasi-optimal
and efficient stationary strategy. If the distribution is unknown, we
additionally show how to sequentially approximate it and, at the same time, act
near-optimally in order to collect as many hidden objects as possible.
| stat |
Pseudo-extended Markov chain Monte Carlo | Sampling from posterior distributions using Markov chain Monte Carlo (MCMC)
methods can require an exhaustive number of iterations, particularly when the
posterior is multi-modal as the MCMC sampler can become trapped in a local mode
for a large number of iterations. In this paper, we introduce the
pseudo-extended MCMC method as a simple approach for improving the mixing of
the MCMC sampler for multi-modal posterior distributions. The pseudo-extended
method augments the state-space of the posterior using pseudo-samples as
auxiliary variables. On the extended space, the modes of the posterior are
connected, which allows the MCMC sampler to easily move between well-separated
posterior modes. We demonstrate that the pseudo-extended approach delivers
improved MCMC sampling over the Hamiltonian Monte Carlo algorithm on
multi-modal posteriors, including Boltzmann machines and models with
sparsity-inducing priors.
| stat |
Assessing the Reliability of Wind Power Operations under a Changing
Climate with a Non-Gaussian Bias Correction | Facing increasing societal and economic pressure, many countries have
established strategies to develop renewable energy portfolios, whose
penetration in the market can alleviate the dependence on fossil fuels. In the
case of wind, there is a fundamental question related to the resilience, and
hence profitability of future wind farms to a changing climate, given that
current wind turbines have lifespans of up to thirty years. In this work, we
develop a new non-Gaussian method data to simulations and to estimate future
wind, predicated on a trans-Gaussian transformation and a cluster-wise
minimization of the Kullback-Leibler divergence. Future winds abundance will be
determined for Saudi Arabia, a country with a recently established plan to
develop a portfolio of up to 16 GW of wind energy. Further, we estimate the
change in profits over future decades using additional high-resolution
simulations, an improved method for vertical wind extrapolation, and power
curves from a collection of popular wind turbines. We find an overall increase
in the daily profit of $272,000 for the wind energy market for the optimal
locations for wind farming in the country.
| stat |
Amortized Inference Regularization | The variational autoencoder (VAE) is a popular model for density estimation
and representation learning. Canonically, the variational principle suggests to
prefer an expressive inference model so that the variational approximation is
accurate. However, it is often overlooked that an overly-expressive inference
model can be detrimental to the test set performance of both the amortized
posterior approximator and, more importantly, the generative density estimator.
In this paper, we leverage the fact that VAEs rely on amortized inference and
propose techniques for amortized inference regularization (AIR) that control
the smoothness of the inference model. We demonstrate that, by applying AIR, it
is possible to improve VAE generalization on both inference and generative
performance. Our paper challenges the belief that amortized inference is simply
a mechanism for approximating maximum likelihood training and illustrates that
regularization of the amortization family provides a new direction for
understanding and improving generalization in VAEs.
| stat |
No-Regret Bayesian Optimization with Unknown Hyperparameters | Bayesian optimization (BO) based on Gaussian process models is a powerful
paradigm to optimize black-box functions that are expensive to evaluate. While
several BO algorithms provably converge to the global optimum of the unknown
function, they assume that the hyperparameters of the kernel are known in
advance. This is not the case in practice and misspecification often causes
these algorithms to converge to poor local optima. In this paper, we present
the first BO algorithm that is provably no-regret and converges to the optimum
without knowledge of the hyperparameters. During optimization we slowly adapt
the hyperparameters of stationary kernels and thereby expand the associated
function class over time, so that the BO algorithm considers more complex
function candidates. Based on the theoretical insights, we propose several
practical algorithms that achieve the empirical sample efficiency of BO with
online hyperparameter estimation, but retain theoretical convergence
guarantees. We evaluate our method on several benchmark problems.
| stat |
Interpreting the Ising Model: The Input Matters | The Ising model is a model for pairwise interactions between binary variables
that has become popular in the psychological sciences. It has been first
introduced as a theoretical model for the alignment between positive (+1) and
negative (-1) atom spins. In many psychological applications, however, the
Ising model is defined on the domain $\{0,1\}$ instead of the classical domain
$\{-1,1\}$. While it is possible to transform the parameters of a given Ising
model in one domain to obtain a statistically equivalent model in the other
domain, the parameters in the two versions of the Ising model lend themselves
to different interpretations and imply different dynamics, when studying the
Ising model as a dynamical system. In this tutorial paper, we provide an
accessible discussion of the interpretation of threshold and interaction
parameters in the two domains and show how the dynamics of the Ising model
depends on the choice of domain. Finally, we provide a transformation that
allows to transform the parameters in an Ising model in one domain into a
statistically equivalent Ising model in the other domain.
| stat |
Within-Person Variability Score-Based Causal Inference: A Two-Step
Semiparametric Estimation for Joint Effects of Time-Varying Treatments | Behavioral science researchers have recently shown strong interest in
disaggregating within- and between-person effects (stable traits) from
longitudinal data. In this paper, we propose a method of within-person
variability score-based causal inference for estimating joint effects of
time-varying continuous treatments by effectively controlling for stable traits
as time-invariant unobserved confounders. After conceptualizing stable trait
factors and within-person variability scores, we introduce the proposed method,
which consists of a two-step analysis. Within-person variability scores for
each person, which are disaggregated from stable traits of that person, are
first calculated using weights based on a best linear correlation preserving
predictor through structural equation modeling. Causal parameters are then
estimated via a potential outcome approach, either marginal structural models
(MSMs) or structural nested mean models (SNMMs), using calculated within-person
variability scores. We emphasize the use of SNMMs with G-estimation because of
its doubly robust property to model errors. Through simulation and empirical
application to data regarding sleep habits and mental health status from the
Tokyo Teen Cohort study, we show that the proposed method can recover causal
parameters well and that causal estimates might be severely biased if one does
not properly account for stable traits.
| stat |
Dynamic Pricing for Airline Ancillaries with Customer Context | Ancillaries have become a major source of revenue and profitability in the
travel industry. Yet, conventional pricing strategies are based on business
rules that are poorly optimized and do not respond to changing market
conditions. This paper describes the dynamic pricing model developed by Deepair
solutions, an AI technology provider for travel suppliers. We present a pricing
model that provides dynamic pricing recommendations specific to each customer
interaction and optimizes expected revenue per customer. The unique nature of
personalized pricing provides the opportunity to search over the market space
to find the optimal price-point of each ancillary for each customer, without
violating customer privacy. In this paper, we present and compare three
approaches for dynamic pricing of ancillaries, with increasing levels of
sophistication: (1) a two-stage forecasting and optimization model using a
logistic mapping function; (2) a two-stage model that uses a deep neural
network for forecasting, coupled with a revenue maximization technique using
discrete exhaustive search; (3) a single-stage end-to-end deep neural network
that recommends the optimal price. We describe the performance of these models
based on both offline and online evaluations. We also measure the real-world
business impact of these approaches by deploying them in an A/B test on an
airline's internet booking website. We show that traditional machine learning
techniques outperform human rule-based approaches in an online setting by
improving conversion by 36% and revenue per offer by 10%. We also provide
results for our offline experiments which show that deep learning algorithms
outperform traditional machine learning techniques for this problem. Our
end-to-end deep learning model is currently being deployed by the airline in
their booking system.
| stat |
A Class of Conjugate Priors for Multinomial Probit Models which Includes
the Multivariate Normal One | Multinomial probit models are widely-implemented representations which allow
both classification and inference by learning changes in vectors of class
probabilities with a set of p observed predictors. Although various frequentist
methods have been developed for estimation, inference and classification within
such a class of models, Bayesian inference is still lagging behind. This is due
to the apparent absence of a tractable class of conjugate priors, that may
facilitate posterior inference on the multinomial probit coefficients. Such an
issue has motivated increasing efforts toward the development of effective
Markov chain Monte Carlo methods, but state-of-the-art solutions still face
severe computational bottlenecks, especially in large p settings. In this
article, we prove that the entire class of unified skew-normal (SUN)
distributions is conjugate to a wide variety of multinomial probit models, and
we exploit the SUN properties to improve upon state-of-art-solutions for
posterior inference and classification both in terms of closed-form results for
key functionals of interest, and also by developing novel computational methods
relying either on independent and identically distributed samples from the
exact posterior or on scalable and accurate variational approximations based on
blocked partially-factorized representations. As illustrated in a
gastrointestinal lesions application, the magnitude of the improvements
relative to current methods is particularly evident, in practice, when the
focus is on large p applications.
| stat |
Skewed Distributions or Transformations? Modelling Skewness for a
Cluster Analysis | Because of its mathematical tractability, the Gaussian mixture model holds a
special place in the literature for clustering and classification. For all its
benefits, however, the Gaussian mixture model poses problems when the data is
skewed or contains outliers. Because of this, methods have been developed over
the years for handling skewed data, and fall into two general categories. The
first is to consider a mixture of more flexible skewed distributions, and the
second is based on incorporating a transformation to near normality. Although
these methods have been compared in their respective papers, there has yet to
be a detailed comparison to determine when one method might be more suitable
than the other. Herein, we provide a detailed comparison on many benchmarking
datasets, as well as describe a novel method to assess cluster separation.
| stat |
The age of secrecy and unfairness in recidivism prediction | In our current society, secret algorithms make important decisions about
individuals. There has been substantial discussion about whether these
algorithms are unfair to groups of individuals. While noble, this pursuit is
complex and ultimately stagnating because there is no clear definition of
fairness and competing definitions are largely incompatible. We argue that the
focus on the question of fairness is misplaced, as these algorithms fail to
meet a more important and yet readily obtainable goal: transparency. As a
result, creators of secret algorithms can provide incomplete or misleading
descriptions about how their models work, and various other kinds of errors can
easily go unnoticed. By partially reverse engineering the COMPAS algorithm -- a
recidivism-risk scoring algorithm used throughout the criminal justice system
-- we show that it does not seem to depend linearly on the defendant's age,
despite statements to the contrary by the algorithm's creator. Furthermore, by
subtracting from COMPAS its (hypothesized) nonlinear age component, we show
that COMPAS does not necessarily depend on race, contradicting ProPublica's
analysis, which assumed linearity in age. In other words, faulty assumptions
about a proprietary algorithm lead to faulty conclusions that go unchecked
without careful reverse engineering. Were the algorithm transparent in the
first place, this would likely not have occurred. The most important result in
this work is that we find that there are many defendants with low risk score
but long criminal histories, suggesting that data inconsistencies occur
frequently in criminal justice databases. We argue that transparency satisfies
a different notion of procedural fairness by providing both the defendants and
the public with the opportunity to scrutinize the methodology and calculations
behind risk scores for recidivism.
| stat |
Robust estimation for small domains in business surveys | Small area (or small domain) estimation is still rarely applied in business
statistics, because of challenges arising from the skewness and variability of
variables such as turnover. We examine a range of small area estimation methods
as the basis for estimating the activity of industries within the retail sector
in the Netherlands. We use tax register data and a sampling procedure which
replicates the sampling for the retail sector of Statistics Netherlands'
Structural Business Survey as a basis for investigating the properties of small
area estimators. In particular, we consider the use of the EBLUP under a random
effects model and variations of the EBLUP derived under (a) a random effects
model that includes a complex specification for the level 1 variance and (b) a
random effects model that is fitted by using the survey weights. Although
accounting for the survey weights in estimation is important, the impact of
influential data points remains the main challenge in this case. The paper
further explores the use of outlier robust estimators in business surveys, in
particular a robust version of the EBLUP, M-regression based synthetic
estimators, and M-quantile small area estimators. The latter family of small
area estimators includes robust projective (without and with survey weights)
and robust predictive versions. M-quantile methods have the lowest empirical
mean squared error and are substantially better than direct estimators, though
there is an open question about how to choose the tuning constant for bias
adjustment in practice. The paper makes a further contribution by exploring a
doubly robust approach comprising the use of survey weights in conjunction with
outlier robust methods in small area estimation.
| stat |
Optimal trees selection for classification via out-of-bag assessment and
sub-bagging | The effect of training data size on machine learning methods has been well
investigated over the past two decades. The predictive performance of tree
based machine learning methods, in general, improves with a decreasing rate as
the size of training data increases. We investigate this in optimal trees
ensemble (OTE) where the method fails to learn from some of the training
observations due to internal validation. Modified tree selection methods are
thus proposed for OTE to cater for the loss of training observations in
internal validation. In the first method, corresponding out-of-bag (OOB)
observations are used in both individual and collective performance assessment
for each tree. Trees are ranked based on their individual performance on the
OOB observations. A certain number of top ranked trees is selected and starting
from the most accurate tree, subsequent trees are added one by one and their
impact is recorded by using the OOB observations left out from the bootstrap
sample taken for the tree being added. A tree is selected if it improves
predictive accuracy of the ensemble. In the second approach, trees are grown on
random subsets, taken without replacement-known as sub-bagging, of the training
data instead of bootstrap samples (taken with replacement). The remaining
observations from each sample are used in both individual and collective
assessments for each corresponding tree similar to the first method. Analysis
on 21 benchmark datasets and simulations studies show improved performance of
the modified methods in comparison to OTE and other state-of-the-art methods.
| stat |
Forecasting the Remittances of the Overseas Filipino Workers in the
Philippines | This study aims to find a Box-Jenkins time series model for the monthly OFW's
remittance in the Philippines. Forecasts of OFW's remittance for the years 2018
and 2019 will be generated using the appropriate time series model. The data
were retrieved from the official website of Bangko Sentral ng Pilipinas. There
are 108 observations, 96 of which were used in model building and the remaining
12 observations were used in forecast evaluation. ACF and PACF were used to
examine the stationarity of the series. Augmented Dickey Fuller test was used
to confirm the stationarity of the series. The data was found to have a
seasonal component, thus, seasonality has been considered in the final model
which is SARIMA (2,1,0)x(0,0,2)_12. There are no significant spikes in the ACF
and PACF of residuals of the final model and the L-jung Box Q* test confirms
further that the residuals of the model are uncorrelated. Also, based on the
result of the Shapiro-Wilk test for the forecast errors, the forecast errors
can be considered a Gaussian white noise. Considering the results of diagnostic
checking and forecast evaluation, SARIMA (2,1,0)x(0,0,2)_12 is an appropriate
model for the series. All necessary computations were done using the R
statistical software.
| stat |
Increasing Power for Observational Studies of Aberrant Response: An
Adaptive Approach | In many observational studies, the interest is in the effect of treatment on
bad, aberrant outcomes rather than the average outcome. For such settings, the
traditional approach is to define a dichotomous outcome indicating aberration
from a continuous score and use the Mantel-Haenszel test with matched data. For
example, studies of determinants of poor child growth use the World Health
Organization's definition of child stunting being height-for-age z-score $\leq
-2$. The traditional approach may lose power because it discards potentially
useful information about the severity of aberration. We develop an adaptive
approach that makes use of this information and asymptotically dominates the
traditional approach. We develop our approach in two parts. First, we develop
an aberrant rank approach in matched observational studies and prove a novel
design sensitivity formula enabling its asymptotic comparison with the
Mantel-Haenszel test under various settings. Second, we develop a new, general
adaptive approach, the two-stage programming method, and use it to adaptively
combine the aberrant rank test and the Mantel-Haenszel test. We apply our
approach to a study of the effect of teenage pregnancy on stunting.
| stat |
FuncNN: An R Package to Fit Deep Neural Networks Using Generalized Input
Spaces | Neural networks have excelled at regression and classification problems when
the input space consists of scalar variables. As a result of this proficiency,
several popular packages have been developed that allow users to easily fit
these kinds of models. However, the methodology has excluded the use of
functional covariates and to date, there exists no software that allows users
to build deep learning models with this generalized input space. To the best of
our knowledge, the functional neural network (FuncNN) library is the first such
package in any programming language; the library has been developed for R and
is built on top of the keras architecture. Throughout this paper, several
functions are introduced that provide users an avenue to easily build models,
generate predictions, and run cross-validations. A summary of the underlying
methodology is also presented. The ultimate contribution is a package that
provides a set of general modelling and diagnostic tools for data problems in
which there exist both functional and scalar covariates.
| stat |
Mixture survival models methodology: an application to cancer
immunotherapy assessment in clinical trials | Progress in immunotherapy revolutionized the treatment landscape for advanced
lung cancer, raising survival expectations beyond those that were historically
anticipated with this disease. In the present study, we describe the methods
for the adjustment of mixture parametric models of two populations for survival
analysis in the presence of long survivors. A methodology is proposed in
several five steps: first, it is proposed to use the multimodality test to
decide the number of subpopulations to be considered in the model, second to
adjust simple parametric survival models and mixture distribution models, to
estimate the parameters and to select the best model fitted the data, finally,
to test the hypotheses to compare the effectiveness of immunotherapies in the
context of randomized clinical trials. The methodology is illustrated with data
from a clinical trial that evaluates the effectiveness of the therapeutic
vaccine CIMAvaxEGF vs the best supportive care for the treatment of advanced
lung cancer. The mixture survival model allows estimating the presence of a
subpopulation of long survivors that is 44% for vaccinated patients. The
differences between the treated and control group were significant in both
subpopulations (population of short-term survival: p = 0.001, the population of
long-term survival: p = 0.0002). For cancer therapies, where a proportion of
patients achieves long-term control of the disease, the heterogeneity of the
population must be taken into account. Mixture parametric models may be more
suitable to detect the effectiveness of immunotherapies compared to standard
models.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.