title
stringlengths 6
244
| abstract
stringlengths 19
5.09k
| label
stringclasses 10
values |
|---|---|---|
Three issues impeding communication of statistical methodology for
incomplete data | We identify three issues permeating the literature on statistical methodology
for incomplete data written for non-specialist statisticians and other
investigators. The first is a mathematical defect in the notation Yobs, Ymis
used to partition the data into observed and missing components. The second are
issues concerning the notation `P(R|Yobs, Ymis)=P(R|Yobs)' used for
communicating the definition of missing at random (MAR). And the third is the
framing of ignorability by emulating complete-data methods exactly, rather than
treating the question of ignorability on its own merits. These issues have been
present in the literature for a long time, and have simple remedies. The
purpose of this paper is to raise awareness of these issues, and to explain how
they can be remedied.
| stat |
Regularization Strategies for Quantile Regression | We investigate different methods for regularizing quantile regression when
predicting either a subset of quantiles or the full inverse CDF. We show that
minimizing an expected pinball loss over a continuous distribution of quantiles
is a good regularizer even when only predicting a specific quantile. For
predicting multiple quantiles, we propose achieving the classic goal of
non-crossing quantiles by using deep lattice networks that treat the quantile
as a monotonic input feature, and we discuss why monotonicity on other features
is an apt regularizer for quantile regression. We show that lattice models
enable regularizing the predicted distribution to a location-scale family.
Lastly, we propose applying rate constraints to improve the calibration of the
quantile predictions on specific subsets of interest and improve fairness
metrics. We demonstrate our contributions on simulations, benchmark datasets,
and real quantile regression problems.
| stat |
Outlier Detection for Functional Data with R Package fdaoutlier | Outlier detection is one of the standard exploratory analysis tasks in
functional data analysis. We present the R package fdaoutlier which contains
implementations of some of the latest techniques for detecting functional
outliers. The package makes it easy to detect different types of outliers
(magnitude, shape, and amplitude) in functional data, and some of the
implemented methods can be applied to both univariate and multivariate
functional data. We illustrate the main functionality of the R package with
common functional datasets in the literature.
| stat |
Towards Optimal Transport with Global Invariances | Many problems in machine learning involve calculating correspondences between
sets of objects, such as point clouds or images. Discrete optimal transport
provides a natural and successful approach to such tasks whenever the two sets
of objects can be represented in the same space, or at least distances between
them can be directly evaluated. Unfortunately neither requirement is likely to
hold when object representations are learned from data. Indeed, automatically
derived representations such as word embeddings are typically fixed only up to
some global transformations, for example, reflection or rotation. As a result,
pairwise distances across two such instances are ill-defined without specifying
their relative transformation. In this work, we propose a general framework for
optimal transport in the presence of latent global transformations. We cast the
problem as a joint optimization over transport couplings and transformations
chosen from a flexible class of invariances, propose algorithms to solve it,
and show promising results in various tasks, including a popular unsupervised
word translation benchmark.
| stat |
Flow Contrastive Estimation of Energy-Based Models | This paper studies a training method to jointly estimate an energy-based
model and a flow-based model, in which the two models are iteratively updated
based on a shared adversarial value function. This joint training method has
the following traits. (1) The update of the energy-based model is based on
noise contrastive estimation, with the flow model serving as a strong noise
distribution. (2) The update of the flow model approximately minimizes the
Jensen-Shannon divergence between the flow model and the data distribution. (3)
Unlike generative adversarial networks (GAN) which estimates an implicit
probability distribution defined by a generator model, our method estimates two
explicit probabilistic distributions on the data. Using the proposed method we
demonstrate a significant improvement on the synthesis quality of the flow
model, and show the effectiveness of unsupervised feature learning by the
learned energy-based model. Furthermore, the proposed training method can be
easily adapted to semi-supervised learning. We achieve competitive results to
the state-of-the-art semi-supervised learning methods.
| stat |
Detecting Multiple Replicating Signals using Adaptive Filtering
Procedures | Replicability is a fundamental quality of scientific discoveries: we are
interested in those signals that are detectable in different laboratories,
study populations, across time etc. Unlike meta-analysis which accounts for
experimental variability but does not guarantee replicability, testing a
partial conjunction (PC) null aims specifically to identify the signals that
are discovered in multiple studies. In many contemporary applications, e.g.,
comparing multiple high-throughput genetic experiments, a large number $M$ of
PC nulls need to be tested simultaneously, calling for a multiple comparisons
correction. However, standard multiple testing adjustments on the $M$ PC
$p$-values can be severely conservative, especially when $M$ is large and the
signals are sparse. We introduce AdaFilter, a new multiple testing procedure
that increases power by adaptively filtering out unlikely candidates of PC
nulls. We prove that AdaFilter can control FWER and FDR as long as data across
studies are independent, and has much higher power than other existing methods.
We illustrate the application of AdaFilter with three examples: microarray
studies of Duchenne muscular dystrophy, single-cell RNA sequencing of T cells
in lung cancer tumors and GWAS for metabolomics.
| stat |
Predictive Uncertainty Quantification with Compound Density Networks | Despite the huge success of deep neural networks (NNs), finding good
mechanisms for quantifying their prediction uncertainty is still an open
problem. Bayesian neural networks are one of the most popular approaches to
uncertainty quantification. On the other hand, it was recently shown that
ensembles of NNs, which belong to the class of mixture models, can be used to
quantify prediction uncertainty. In this paper, we build upon these two
approaches. First, we increase the mixture model's flexibility by replacing the
fixed mixing weights by an adaptive, input-dependent distribution (specifying
the probability of each component) represented by NNs, and by considering
uncountably many mixture components. The resulting class of models can be seen
as the continuous counterpart to mixture density networks and is therefore
referred to as compound density networks (CDNs). We employ both maximum
likelihood and variational Bayesian inference to train CDNs, and empirically
show that they yield better uncertainty estimates on out-of-distribution data
and are more robust to adversarial examples than the previous approaches.
| stat |
The Gaussian Neural Process | Neural Processes (NPs; Garnelo et al., 2018a,b) are a rich class of models
for meta-learning that map data sets directly to predictive stochastic
processes. We provide a rigorous analysis of the standard maximum-likelihood
objective used to train conditional NPs. Moreover, we propose a new member to
the Neural Process family called the Gaussian Neural Process (GNP), which
models predictive correlations, incorporates translation equivariance, provides
universal approximation guarantees, and demonstrates encouraging performance.
| stat |
Interpretable feature subset selection: A Shapley value based approach | For feature selection and related problems, we introduce the notion of
classification game, a cooperative game, with features as players and hinge
loss based characteristic function and relate a feature's contribution to
Shapley value based error apportioning (SVEA) of total training error. Our
major contribution is ($\star$) to show that for any dataset the threshold 0 on
SVEA value identifies feature subset whose joint interactions for label
prediction is significant or those features that span a subspace where the data
is predominantly lying. In addition, our scheme ($\star$) identifies the
features on which Bayes classifier doesn't depend but any surrogate loss
function based finite sample classifier does; this contributes to the excess
$0$-$1$ risk of such a classifier, ($\star$) estimates unknown true hinge risk
of a feature, and ($\star$) relate the stability property of an allocation and
negative valued SVEA by designing the analogue of core of classification game.
Due to Shapley value's computationally expensive nature, we build on a known
Monte Carlo based approximation algorithm that computes characteristic function
(Linear Programs) only when needed. We address the potential sample bias
problem in feature selection by providing interval estimates for SVEA values
obtained from multiple sub-samples. We illustrate all the above aspects on
various synthetic and real datasets and show that our scheme achieves better
results than existing recursive feature elimination technique and ReliefF in
most cases. Our theoretically grounded classification game in terms of well
defined characteristic function offers interpretability (which we formalize in
terms of final task) and explainability of our framework, including
identification of important features.
| stat |
A General Framework for the Derandomization of PAC-Bayesian Bounds | PAC-Bayesian bounds are known to be tight and informative when studying the
generalization ability of randomized classifiers. However, when applied to some
family of deterministic models such as neural networks, they require a loose
and costly derandomization step. As an alternative to this step, we introduce
three new PAC-Bayesian generalization bounds that have the originality to be
pointwise, meaning that they provide guarantees over one single hypothesis
instead of the usual averaged analysis. Our bounds are rather general,
potentially parameterizable, and provide novel insights for various machine
learning settings that rely on randomized algorithms. We illustrate the
interest of our theoretical result for the analysis of neural network training.
| stat |
Generalization and Memorization: The Bias Potential Model | Models for learning probability distributions such as generative models and
density estimators behave quite differently from models for learning functions.
One example is found in the memorization phenomenon, namely the ultimate
convergence to the empirical distribution, that occurs in generative
adversarial networks (GANs). For this reason, the issue of generalization is
more subtle than that for supervised learning. For the bias potential model, we
show that dimension-independent generalization accuracy is achievable if early
stopping is adopted, despite that in the long term, the model either memorizes
the samples or diverges.
| stat |
The sharp, the flat and the shallow: Can weakly interacting agents learn
to escape bad minima? | An open problem in machine learning is whether flat minima generalize better
and how to compute such minima efficiently. This is a very challenging problem.
As a first step towards understanding this question we formalize it as an
optimization problem with weakly interacting agents. We review appropriate
background material from the theory of stochastic processes and provide
insights that are relevant to practitioners. We propose an algorithmic
framework for an extended stochastic gradient Langevin dynamics and illustrate
its potential. The paper is written as a tutorial, and presents an alternative
use of multi-agent learning. Our primary focus is on the design of algorithms
for machine learning applications; however the underlying mathematical
framework is suitable for the understanding of large scale systems of agent
based models that are popular in the social sciences, economics and finance.
| stat |
Hierarchical Causal Bandit | Causal bandit is a nascent learning model where an agent sequentially
experiments in a causal network of variables, in order to identify the
reward-maximizing intervention. Despite the model's wide applicability,
existing analytical results are largely restricted to a parallel bandit version
where all variables are mutually independent. We introduce in this work the
hierarchical causal bandit model as a viable path towards understanding general
causal bandits with dependent variables. The core idea is to incorporate a
contextual variable that captures the interaction among all variables with
direct effects. Using this hierarchical framework, we derive sharp insights on
algorithmic design in causal bandits with dependent arms and obtain nearly
matching regret bounds in the case of a binary context.
| stat |
Low-rank Tensor Estimation via Riemannian Gauss-Newton: Statistical
Optimality and Second-Order Convergence | In this paper, we consider the estimation of a low Tucker rank tensor from a
number of noisy linear measurements. The general problem covers many specific
examples arising from applications, including tensor regression, tensor
completion, and tensor PCA/SVD. We propose a Riemannian Gauss-Newton (RGN)
method with fast implementations for low Tucker rank tensor estimation.
Different from the generic (super)linear convergence guarantee of RGN in the
literature, we prove the first quadratic convergence guarantee of RGN for
low-rank tensor estimation under some mild conditions. A deterministic
estimation error lower bound, which matches the upper bound, is provided that
demonstrates the statistical optimality of RGN. The merit of RGN is illustrated
through two machine learning applications: tensor regression and tensor SVD.
Finally, we provide the simulation results to corroborate our theoretical
findings.
| stat |
Augmented Ensemble MCMC sampling in Factorial Hidden Markov Models | Bayesian inference for factorial hidden Markov models is challenging due to
the exponentially sized latent variable space. Standard Monte Carlo samplers
can have difficulties effectively exploring the posterior landscape and are
often restricted to exploration around localised regions that depend on
initialisation. We introduce a general purpose ensemble Markov Chain Monte
Carlo (MCMC) technique to improve on existing poorly mixing samplers. This is
achieved by combining parallel tempering and an auxiliary variable scheme to
exchange information between the chains in an efficient way. The latter
exploits a genetic algorithm within an augmented Gibbs sampler. We compare our
technique with various existing samplers in a simulation study as well as in a
cancer genomics application, demonstrating the improvements obtained by our
augmented ensemble approach.
| stat |
Density Estimation on a Network | This paper develops a novel approach to density estimation on a network. We
formulate nonparametric density estimation on a network as a nonparametric
regression problem by binning. Nonparametric regression using local polynomial
kernel-weighted least squares have been studied rigorously, and its asymptotic
properties make it superior to kernel estimators such as the Nadaraya-Watson
estimator. When applied to a network, the best estimator near a vertex depends
on the amount of smoothness at the vertex. Often, there are no compelling
reasons to assume that a density will be continuous or discontinuous at a
vertex, hence a data driven approach is proposed. To estimate the density in a
neighborhood of a vertex, we propose a two-step procedure. The first step of
this pretest estimator fits a separate local polynomial regression on each edge
using data only on that edge, and then tests for equality of the estimates at
the vertex. If the null hypothesis is not rejected, then the second step
re-estimates the regression function in a small neighborhood of the vertex,
subject to a joint equality constraint. Since the derivative of the density may
be discontinuous at the vertex, we propose a piecewise polynomial local
regression estimate to model the change in slope. We study in detail the
special case of local piecewise linear regression and derive the leading bias
and variance terms using weighted least squares theory. We show that the
proposed approach will remove the bias near a vertex that has been noted for
existing methods, which typically do not allow for discontinuity at vertices.
For a fixed network, the proposed method scales sub-linearly with sample size
and it can be extended to regression and varying coefficient models on a
network. We demonstrate the workings of the proposed model by simulation
studies and apply it to a dendrite network data set.
| stat |
"Old Techniques for New Times": the RMaCzek package for producing
Czekanowski's Diagrams | Inspired by the MaCzek Visual Basic program we provide an R package, RMaCzek,
that produces Czekanowski's diagram. Our package permits any seriation and
distance method the user provides. In this paper we focus on the OLO and
QAP_2SUM methods from the seriation package. We illustrate the possibilities of
our package with three anthropological studies, one socio-economical one and a
phylogenetically motivated simulation study.
| stat |
Doubly Robust Nonparametric Instrumental Variable Estimators for
Survival Outcomes | Instrumental variable (IV) methods allow us the opportunity to address
unmeasured confounding in causal inference. However, most IV methods are only
applicable to discrete or continuous outcomes with very few IV methods for
censored survival outcomes. In this work we propose nonparametric estimators
for the local average treatment effect on survival probabilities under both
nonignorable and ignorable censoring. We provide an efficient influence
function-based estimator and a simple estimation procedure when the IV is
either binary or continuous. The proposed estimators possess double-robustness
properties and can easily incorporate nonparametric estimation using machine
learning tools. In simulation studies, we demonstrate the flexibility and
efficiency of our proposed estimators under various plausible scenarios. We
apply our method to the Prostate, Lung, Colorectal, and Ovarian Cancer
Screening Trial for estimating the causal effect of screening on survival
probabilities and investigate the causal contrasts between the two
interventions under different censoring assumptions.
| stat |
Asymptotically optimal sequential FDR and pFDR control with (or without)
prior information on the number of signals | We investigate asymptotically optimal multiple testing procedures for streams
of sequential data in the context of prior information on the number of false
null hypotheses ("signals"). We show that the "gap" and "gap-intersection"
procedures, recently proposed and shown by Song and Fellouris (2017, Electron.
J. Statist.) to be asymptotically optimal for controlling type 1 and 2
familywise error rates (FWEs), are also asymptotically optimal for controlling
FDR/FNR when their critical values are appropriately adjusted. Generalizing
this result, we show that these procedures, again with appropriately adjusted
critical values, are asymptotically optimal for controlling any multiple
testing error metric that is bounded between multiples of FWE in a certain
sense. This class of metrics includes FDR/FNR but also pFDR/pFNR, the
per-comparison and per-family error rates, and the false positive rate. Our
analysis includes asymptotic regimes in which the number of null hypotheses
approaches $\infty$ as the type 1 and 2 error metrics approach $0$.
| stat |
When stakes are high: balancing accuracy and transparency with
Model-Agnostic Interpretable Data-driven suRRogates | Highly regulated industries, like banking and insurance, ask for transparent
decision-making algorithms. At the same time, competitive markets are pushing
for the use of complex black box models. We therefore present a procedure to
develop a Model-Agnostic Interpretable Data-driven suRRogate (maidrr) suited
for structured tabular data. Knowledge is extracted from a black box via
partial dependence effects. These are used to perform smart feature engineering
by grouping variable values. This results in a segmentation of the feature
space with automatic variable selection. A transparent generalized linear model
(GLM) is fit to the features in categorical format and their relevant
interactions. We demonstrate our R package maidrr with a case study on general
insurance claim frequency modeling for six publicly available datasets. Our
maidrr GLM closely approximates a gradient boosting machine (GBM) black box and
outperforms both a linear and tree surrogate as benchmarks.
| stat |
Spatial Analysis Made Easy with Linear Regression and Kernels | Kernel methods are an incredibly popular technique for extending linear
models to non-linear problems via a mapping to an implicit, high-dimensional
feature space. While kernel methods are computationally cheaper than an
explicit feature mapping, they are still subject to cubic cost on the number of
points. Given only a few thousand locations, this computational cost rapidly
outstrips the currently available computational power. This paper aims to
provide an overview of kernel methods from first-principals (with a focus on
ridge regression), before progressing to a review of random Fourier features
(RFF), a set of methods that enable the scaling of kernel methods to big
datasets. At each stage, the associated R code is provided. We begin by
illustrating how the dual representation of ridge regression relies solely on
inner products and permits the use of kernels to map the data into
high-dimensional spaces. We progress to RFFs, showing how only a few lines of
code provides a significant computational speed-up for a negligible cost to
accuracy. We provide an example of the implementation of RFFs on a simulated
spatial data set to illustrate these properties. Lastly, we summarise the main
issues with RFFs and highlight some of the advanced techniques aimed at
alleviating them.
| stat |
A Robust Statistical method to Estimate the Intervention Effect with
Longitudinal Data | Segmented regression is a standard statistical procedure used to estimate the
effect of a policy intervention on time series outcomes. This statistical
method assumes the normality of the outcome variable, a large sample size, no
autocorrelation in the observations, and a linear trend over time. Also,
segmented regression is very sensitive to outliers. In a small sample study, if
the outcome variable does not follow a Gaussian distribution, then using
segmented regression to estimate the intervention effect leads to incorrect
inferences. To address the small sample problem and non-normality in the
outcome variable, including outliers, we describe and develop a robust
statistical method to estimate the policy intervention effect in a series of
longitudinal data. A simulation study is conducted to demonstrate the effect of
outliers and non-normality in the outcomes by calculating the power of the test
statistics with the segmented regression and the proposed robust statistical
methods. Moreover, since finding the sampling distribution of the proposed
robust statistic is analytically difficult, we use a nonparametric bootstrap
technique to study the properties of the sampling distribution and make
statistical inferences. Simulation studies show that the proposed method has
more power than the standard t-test used in segmented regression analysis under
the non-normality error distribution. Finally, we use the developed technique
to estimate the intervention effect of the Istanbul Declaration on illegal
organ activities. The robust method detected more significant effects compared
to the standard method and provided shorter confidence intervals.
| stat |
Bayesian Optimization for Materials Design with Mixed Quantitative and
Qualitative Variables | Although Bayesian Optimization (BO) has been employed for accelerating
materials design in computational materials engineering, existing works are
restricted to problems with quantitative variables. However, real designs of
materials systems involve both qualitative and quantitative design variables
representing material compositions, microstructure morphology, and processing
conditions. For mixed-variable problems, existing Bayesian Optimization (BO)
approaches represent qualitative factors by dummy variables first and then fit
a standard Gaussian process (GP) model with numerical variables as the
surrogate model. This approach is restrictive theoretically and fails to
capture complex correlations between qualitative levels. We present in this
paper the integration of a novel latent-variable (LV) approach for
mixed-variable GP modeling with the BO framework for materials design. LVGP is
a fundamentally different approach that maps qualitative design variables to
underlying numerical LV in GP, which has strong physical justification. It
provides flexible parameterization and representation of qualitative factors
and shows superior modeling accuracy compared to the existing methods. We
demonstrate our approach through testing with numerical examples and materials
design examples. It is found that in all test examples the mapped LVs provide
intuitive visualization and substantial insight into the nature and effects of
the qualitative factors. Though materials designs are used as examples, the
method presented is generic and can be utilized for other mixed variable design
optimization problems that involve expensive physics-based simulations.
| stat |
Bayesian Fusion of Data Partitioned Particle Estimates | We present a Bayesian data fusion method to approximate a posterior
distribution from an ensemble of particle estimates that only have access to
subsets of the data. Our approach relies on approximate probabilistic inference
of model parameters through Monte Carlo methods, followed by an update and
resample scheme related to multiple importance sampling to combine information
from the initial estimates. We show the method is convergent in the particle
limit and directly suited to application on multi-sensor data fusion problems
by demonstrating efficacy on a multi-sensor Keplerian orbit determination
problem and a bearings-only tracking problem.
| stat |
Histopathological imaging features- versus molecular measurements-based
cancer prognosis modeling | For most if not all cancers, prognosis is of significant importance, and
extensive modeling research has been conducted. With the genetic nature of
cancer, in the past two decades, multiple types of molecular data (such as gene
expressions and DNA mutations) have been explored. More recently,
histopathological imaging data, which is routinely collected in biopsy, has
been shown as informative for modeling prognosis. In this study, using the TCGA
LUAD and LUSC data as a showcase, we examine and compare modeling lung cancer
overall survival using gene expressions versus histopathological imaging
features. High-dimensional regularization methods are adopted for estimation
and selection. Our analysis shows that gene expressions have slightly better
prognostic performance. In addition, most of the gene expressions are found to
be weakly correlated imaging features. It is expected that this study can
provide some insight into utilizing the two types of important data in cancer
prognosis modeling and into lung cancer overall survival.
| stat |
Variational Integrator Networks for Physically Structured Embeddings | Learning workable representations of dynamical systems is becoming an
increasingly important problem in a number of application areas. By leveraging
recent work connecting deep neural networks to systems of differential
equations, we propose \emph{variational integrator networks}, a class of neural
network architectures designed to preserve the geometric structure of physical
systems. This class of network architectures facilitates accurate long-term
prediction, interpretability, and data-efficient learning, while still
remaining highly flexible and capable of modeling complex behavior. We
demonstrate that they can accurately learn dynamical systems from both noisy
observations in phase space and from image pixels within which the unknown
dynamics are embedded.
| stat |
Maximum $\log_q$ Likelihood Estimation for Parameters of Weibull
Distribution and Properties: Monte Carlo Simulation | The maximum ${\log}_q$ likelihood estimation method is a generalization of
the known maximum $\log$ likelihood method to overcome the problem for modeling
non-identical observations (inliers and outliers). The parameter $q$ is a
tuning constant to manage the modeling capability. Weibull is a flexible and
popular distribution for problems in engineering. In this study, this method is
used to estimate the parameters of Weibull distribution when non-identical
observations exist. Since the main idea is based on modeling capability of
objective function
$\rho(x;\boldsymbol{\theta})=\log_q\big[f(x;\boldsymbol{\theta})\big]$, we
observe that the finiteness of score functions cannot play a role in the robust
estimation for inliers. The properties of Weibull distribution are examined. In
the numerical experiment, the parameters of Weibull distribution are estimated
by $\log_q$ and its special form, $\log$, likelihood methods if the different
designs of contamination into underlying Weibull distribution are applied. The
optimization is performed via genetic algorithm. The modeling competence of
$\rho(x;\boldsymbol{\theta})$ and insensitiveness to non-identical observations
are observed by Monte Carlo simulation. The value of $q$ can be chosen by use
of the mean squared error in simulation and the $p$-value of Kolmogorov-Smirnov
test statistic used for evaluation of fitting competence. Thus, we can overcome
the problem about determining of the value of $q$ for real data sets.
| stat |
Implicit bias of gradient descent for mean squared error regression with
wide neural networks | We investigate gradient descent training of wide neural networks and the
corresponding implicit bias in function space. Focusing on 1D regression, we
show that the solution of training a width-$n$ shallow ReLU network is within
$n^{- 1/2}$ of the function which fits the training data and whose difference
from initialization has smallest 2-norm of the second derivative weighted by
$1/\zeta$. The curvature penalty function $1/\zeta$ is expressed in terms of
the probability distribution that is utilized to initialize the network
parameters, and we compute it explicitly for various common initialization
procedures. For instance, asymmetric initialization with a uniform distribution
yields a constant curvature penalty, and thence the solution function is the
natural cubic spline interpolation of the training data. The statement
generalizes to the training trajectories, which in turn are captured by
trajectories of spatially adaptive smoothing splines with decreasing
regularization strength.
| stat |
Laplace approximation for fast Bayesian inference in generalized
additive models based on penalized regression splines | Generalized additive models (GAMs) are a well-established statistical tool
for modeling complex nonlinear relationships between covariates and a response
assumed to have a conditional distribution in the exponential family. In this
article, P-splines and the Laplace approximation are coupled for flexible and
fast approximate Bayesian inference in GAMs. The proposed Laplace-P-spline
model contributes to the development of a new methodology to explore the
posterior penalty space by considering a deterministic grid-based strategy or a
Markov chain sampler, depending on the number of smooth additive terms in the
predictor. Our approach has the merit of relying on closed form analytical
expressions for the gradient and Hessian of the approximate posterior penalty
vector, which enables to construct accurate posterior pointwise and credible
set estimators for latent field variables at a relatively low computational
budget even for a large number of smooth additive components. Based upon simple
Gaussian approximations of the conditional latent field posterior, the
suggested methodology enjoys excellent statistical properties. The performance
of the Laplace-P-spline model is confirmed through different simulation
scenarios and the method is illustrated on two real datasets.
| stat |
Incorporating Open Data into Introductory Courses in Statistics | The 2016 Guidelines for Assessment and Instruction in Statistics Education
(GAISE) College Report emphasized six recommendations to teach introductory
courses in statistics. Among them: use of real data with context and purpose.
Many educators have created databases consisting of multiple data sets for use
in class; sometimes making hundreds of data sets available. Yet `the context
and purpose' component of the data may remain elusive if just a generic
database is made available.
We describe the use of open data in introductory courses. Countries and
cities continue to share data through open data portals. Hence, educators can
find regional data that engages their students more effectively. We present
excerpts from case studies that show the application of statistical methods to
data on: crime, housing, rainfall, tourist travel, and others. Data wrangling
and discussion of results are recognized as important case study components.
Thus the open data based case studies attend most GAISE College Report
recommendations. Reproducible \textsf{R} code is made available for each case
study. Example uses of open data in more advanced courses in statistics are
also described.
| stat |
Generalizing the log-Moyal distribution and regression models for heavy
tailed loss data | Catastrophic loss data are known to be heavy-tailed. Practitioners then need
models that are able to capture both tail and modal parts of claim data. To
this purpose, a new parametric family of loss distributions is proposed as a
gamma mixture of the generalized log-Moyal distribution from Bhati and Ravi
(2018), termed the generalized log-Moyal gamma distribution (GLMGA). We discuss
the probabilistic characteristics of the GLMGA, and statistical estimation of
the parameters through maximum likelihood. While the GLMGA distribution is a
special case of the GB2 distribution, we show that this simpler model is
effective in regression modelling of large and modal loss data. A fire claim
data set reported in Cummins et al. (1990) and a Chinese earthquake loss data
set are used to illustrate the applicability of the proposed model.
| stat |
The effect of short-term exposure to the natural environment on
depressive mood: A systematic review and meta-analysis | Research suggests that exposure to the natural environment can improve mood,
however, current reviews are limited in scope and there is little understanding
of moderators. We aimed to conduct a comprehensive systematic review and
meta-analysis of the evidence for the effect of short-term exposure to the
natural environment on depressive mood. Five databases were systematically
searched for relevant studies published up to March 2018. Risk of bias was
evaluated using the Cochrane Risk of Bias (ROB) tool 1.0 and the Risk of Bias
in Non-Randomised Studies of Interventions (ROBINS-I) tool where appropriate.
The GRADE approach was used to assess the quality of evidence overall. A
random-effects meta-analysis was performed. 20 potential moderators of the
effect size were coded and the machine learning-based MetaForest algorithm was
used to identify relevant moderators. These were then entered into a
meta-regression. 33 studies met the inclusion criteria. Effect sizes ranged
from -2.30 to 0.84, with a pooled effect size of $\gamma$ = -0.30 95% CI [-0.50
to -0.10]. However, there was significant residual heterogeneity between
studies and risk of bias was high. Type of natural environment, type of built
environment, gender mix of the sample, and region of study origin, among
others, were identified as relevant moderators but were not significant when
entered in a meta-regression. Quality of evidence was rated very low to low. An
assessment of publication bias was inconclusive. A small effect was found for
reduction in depressive mood following exposure to the natural environment.
However, the high risk of bias and low quality of studies limits confidence in
the results. The variation in effect size also remains largely unexplained. It
is recommended that future studies make use of reporting guidelines and aim to
reduce the potential for bias where possible.
| stat |
OmicsMapNet: Transforming omics data to take advantage of Deep
Convolutional Neural Network for discovery | We developed OmicsMapNet approach to take advantage of existing deep leaning
frameworks to analyze high-dimensional omics data as 2-dimensional images. The
omics data of individual samples were first rearranged into 2D images in which
molecular features related in functions, ontologies, or other relationships
were organized in spatially adjacent and patterned locations. Deep learning
neural networks were trained to classify the images. Molecular features
informative of classes of different phenotypes were subsequently identified. As
an example, we used the KEGG BRITE database to rearrange RNA-Seq expression
data of TCGA diffuse glioma samples as treemaps to capture the functional
hierarchical structure of genes in 2D images. Deep Convolutional Neural
Networks (CNN) were derived using tools from TensorFlow to learn the grade of
TCGA LGG and GBM samples with relatively high accuracy. The most contributory
features in the trained CNN were confirmed in pathway analysis for their
plausible functional involvement.
| stat |
Semi-supervised Logistic Learning Based on Exponential Tilt Mixture
Models | Consider semi-supervised learning for classification, where both labeled and
unlabeled data are available for training. The goal is to exploit both datasets
to achieve higher prediction accuracy than just using labeled data alone. We
develop a semi-supervised logistic learning method based on exponential tilt
mixture models, by extending a statistical equivalence between logistic
regression and exponential tilt modeling. We study maximum nonparametric
likelihood estimation and derive novel objective functions which are shown to
be Fisher consistent. We also propose regularized estimation and construct
simple and highly interpretable EM algorithms. Finally, we present numerical
results which demonstrate the advantage of the proposed methods compared with
existing methods.
| stat |
Robust estimation of tree structured Gaussian Graphical Model | Consider jointly Gaussian random variables whose conditional independence
structure is specified by a graphical model. If we observe realizations of the
variables, we can compute the covariance matrix, and it is well known that the
support of the inverse covariance matrix corresponds to the edges of the
graphical model. Instead, suppose we only have noisy observations. If the noise
at each node is independent, we can compute the sum of the covariance matrix
and an unknown diagonal. The inverse of this sum is (in general) dense. We ask:
can the original independence structure be recovered? We address this question
for tree structured graphical models. We prove that this problem is
unidentifiable, but show that this unidentifiability is limited to a small
class of candidate trees. We further present additional constraints under which
the problem is identifiable. Finally, we provide an O(n^3) algorithm to find
this equivalence class of trees.
| stat |
Going Deep: Models for Continuous-Time Within-Play Valuation of Game
Outcomes in American Football with Tracking Data | Continuous-time assessments of game outcomes in sports have become
increasingly common in the last decade. In American football, only
discrete-time estimates of play value were possible, since the most advanced
public football datasets were recorded at the play-by-play level. While
measures such as expected points and win probability are useful for evaluating
football plays and game situations, there has been no research into how these
values change throughout the course of a play. In this work, we make two main
contributions: First, we introduce a general framework for continuous-time
within-play valuation in the National Football League using player-tracking
data. Our modular framework incorporates several modular sub-models, to easily
incorporate recent work involving player tracking data in football. Second, we
use a long short-term memory recurrent neural network to construct a
ball-carrier model to estimate how many yards the ball-carrier is expected to
gain from their current position, conditional on the locations and trajectories
of the ball-carrier, their teammates and opponents. Additionally, we
demonstrate an extension with conditional density estimation so that the
expectation of any measure of play value can be calculated in continuous-time,
which was never before possible at such a granular level.
| stat |
Threshold selection for wave heights: asymptotic methods based on
L-moments | Two automatic threshold selection (TS) methods for Extreme Value analysis
under a peaks-over-threshold (POT) approach are presented and evaluated, both
built on: fitting the Generalized Pareto distribution (GPd) to excesses'
samples over candidate levels ; the GPd-specific relation between L-skewness
and L-kurtosis; the asymptotic behaviour of the matching L-statistics.
Performance is illustrated on significant wave heights data sets and compared
to the L-moment-based heuristic in [10], which is found to be favorable.
PUBLISHED VERSION AVAILABLE AT:
https://www.spestatistica.pt/storage/app/uploads/public/609/28f/6d0/60928f6d08a0c016386627.pdf
| stat |
A comparison of parameter estimation in function-on-function regression | Recent technological developments have enabled us to collect complex and
high-dimensional data in many scientific fields, such as population health,
meteorology, econometrics, geology, and psychology. It is common to encounter
such datasets collected repeatedly over a continuum. Functional data, whose
sample elements are functions in the graphical forms of curves, images, and
shapes, characterize these data types. Functional data analysis techniques
reduce the complex structure of these data and focus on the dependences within
and (possibly) between the curves. A common research question is to investigate
the relationships in regression models that involve at least one functional
variable. However, the performance of functional regression models depends on
several factors, such as the smoothing technique, the number of basis
functions, and the estimation method. This paper provides a selective
comparison for function-on-function regression models where both the response
and predictor(s) are functions, to determine the optimal choice of basis
function from a set of model evaluation criteria. We also propose a bootstrap
method to construct a confidence interval for the response function. The
numerical comparisons are implemented through Monte Carlo simulations and two
real data examples.
| stat |
Bayesian nonparametric multiway regression for clustered binomial data | We introduce a Bayesian nonparametric regression model for data with multiway
(tensor) structure, motivated by an application to periodontal disease (PD)
data. Our outcome is the number of diseased sites measured over four different
tooth types for each subject, with subject-specific covariates available as
predictors. The outcomes are not well-characterized by simple parametric
models, so we use a nonparametric approach with a binomial likelihood wherein
the latent probabilities are drawn from a mixture with an arbitrary number of
components, analogous to a Dirichlet Process (DP). We use a flexible probit
stick-breaking formulation for the component weights that allows for covariate
dependence and clustering structure in the outcomes. The parameter space for
this model is large and multiway: patients $\times$ tooth types $\times$
covariates $\times$ components. We reduce its effective dimensionality, and
account for the multiway structure, via low-rank assumptions. We illustrate how
this can improve performance, and simplify interpretation, while still
providing sufficient flexibility. We describe a general and efficient Gibbs
sampling algorithm for posterior computation. The resulting fit to the PD data
outperforms competitors, and is interpretable and well-calibrated. An
interactive visual of the predictive model is available at
http://ericfrazerlock.com/toothdata/ToothDisplay.html , and the code is
available at https://github.com/lockEF/NonparametricMultiway .
| stat |
Enforcing Mean Reversion in State Space Models for Prawn Pond Water
Quality Forecasting | The contribution of this study is a novel approach to introduce mean
reversion in multi-step-ahead forecasts of state-space models. This approach is
demonstrated in a prawn pond water quality forecasting application. The mean
reversion constrains forecasts by gradually drawing them to an average of
previously observed dynamics. This corrects deviations in forecasts caused by
irregularities such as chaotic, non-linear, and stochastic trends. The key
features of the approach include (1) it enforces mean reversion, (2) it
provides a means to model both short and long-term dynamics, (3) it is able to
apply mean reversion to select structural state-space components, and (4) it is
simple to implement. Our mean reversion approach is demonstrated on various
state-space models and compared with several time-series models on a prawn pond
water quality dataset. Results show that mean reversion reduces long-term
forecast errors by over 60% to produce the most accurate models in the
comparison.
| stat |
Hierarchical Bayesian propulsion power models for marine vessels | Assessing the magnitude of fuel consumption of marine traffic is a
challenging task. The consumption can be reduced by the ways the vessels are
operated, to achieve both improved cost efficiency and reduced CO2 emissions.
Mathematical models for predicting ships' consumption are in a central role in
both of these tasks. Nowadays, many ships are equipped with data collection
systems, which enable data-based calibration of the consumption models.
Typically this calibration procedure is carried out independently for each
particular ship, using only data collected from the ship in question. In this
paper, we demonstrate a hierarchical Bayesian modeling approach, where we fit a
single model over many vessels, with the assumption that the parameters of
vessels of same type and similar characteristics (e.g. vessel size) are likely
close to each other. The benefits of such an approach are two-fold; 1) we can
borrow information about parameters that are not well informed by the
vessel-specific data using data from similar ships, and 2) we can use the final
hierarchical model to predict the behavior of a vessel from which we don't have
any data, based only on its characteristics. In this paper, we discuss the
basic concept and present a first simple version of the model. We apply the
Stan statistical modeling tool for the model fitting and use real data from 64
cruise ships collected via the widely used commercial Eniram platform. By using
Bayesian statistical methods we obtain uncertainties for the model predictions,
too. The prediction accuracy of the model is compared to an existing data-free
modeling approach.
| stat |
Modifying the Chi-square and the CMH test for population genetic
inference: adapting to over-dispersion | Evolve and resequence studies provide a popular approach to simulate
evolution in the lab and explore its genetic basis. In this context, the
chi-square test, Fishers exact test, as well as the Cochran-Mantel-Haenszel
test are commonly used to infer genomic positions affected by selection from
temporal changes in allele frequency. However, the null model associated with
these tests does not match the null hypothesis of actual interest. Indeed due
to genetic drift and possibly other additional noise components such as pool
sequencing, the null variance in the data can be substantially larger than
accounted forby these common test statistics. This leads to p-values that are
systematically too small and therefore a huge number of false positive results.
Even, if the ranking rather than the actual p-values is of interest, a naive
application of the mentioned tests will give misleading results, as the amount
of over-dispersion varies from locus to locus. We therefore propose adjusted
statistics that take the over-dispersion into account while keeping the
formulas simple. This is particularly useful in genome-wide applications, where
millions of SNPs can be handled with little computational effort. We then apply
the adapted test statistics to real data fromDrosophila, and investigate how
in-formation from intermediate generations can be included when avail-able. The
obtained formulas may also be useful in other situations, provided that the
null variance either is known or can be estimated.
| stat |
Efficient Integration of Aggregate Data and Individual Patient Data in
One-Way Mixed Models | Often both Aggregate Data (AD) studies and Individual Patient Data (IPD)
studies are available for specific treatments. Combining these two sources of
data could improve the overall meta-analytic estimates of treatment effects.
Moreover, often for some studies with AD, the associated IPD maybe available,
albeit at some extra effort or cost to the analyst. We propose a method for
combining treatment effects across trials when the response is from the
exponential family of distribution and hence a generalized linear model
structure can be used. We consider the case when treatment effects are fixed
and common across studies. Using the proposed combination method, we evaluate
the wisdom of choosing AD when IPD is available by studying the relative
efficiency of analyzing all IPD studies versus combining various percentages of
AD and IPD studies. For many different models design constraints under which
the AD estimators are the IPD estimators, and hence fully efficient, are known.
For such models we advocate a selection procedure that chooses AD studies over
IPD studies in a manner that force least departure from design constraints and
hence ensures a fully efficient combined AD and IPD estimator.
| stat |
SONIC: SOcial Network with Influencers and Communities | The integration of social media characteristics into an econometric framework
requires modeling a high dimensional dynamic network with dimensions of
parameter typically much larger than the number of observations. To cope with
this problem, we introduce SONIC, a new high-dimensional network model that
assumes that (1) only few influencers drive the network dynamics; (2) the
community structure of the network is characterized by homogeneity of response
to specific influencers, implying their underlying similarity. An estimation
procedure is proposed based on a greedy algorithm and LASSO regularization.
Through theoretical study and simulations, we show that the matrix parameter
can be estimated even when sample size is smaller than the size of the network.
Using a novel dataset retrieved from one of leading social media platforms -
StockTwits and quantifying their opinions via natural language processing, we
model the opinions network dynamics among a select group of users and further
detect the latent communities. With a sparsity regularization, we can identify
important nodes in the network.
| stat |
Nonparametric graphical model for counts | Although multivariate count data are routinely collected in many application
areas, there is surprisingly little work developing flexible models for
characterizing their dependence structure. This is particularly true when
interest focuses on inferring the conditional independence graph. In this
article, we propose a new class of pairwise Markov random field-type models for
the joint distribution of a multivariate count vector. By employing a novel
type of transformation, we avoid restricting to non-negative dependence
structures or inducing other restrictions through truncations. Taking a
Bayesian approach to inference, we choose a Dirichlet process prior for the
distribution of a random effect to induce great flexibility in the
specification. An efficient Markov chain Monte Carlo (MCMC) algorithm is
developed for posterior computation. We prove various theoretical properties,
including posterior consistency, and show that our COunt Nonparametric
Graphical Analysis (CONGA) approach has good performance relative to
competitors in simulation studies. The methods are motivated by an application
to neuron spike count data in mice.
| stat |
Full Law Identification In Graphical Models Of Missing Data:
Completeness Results | Missing data has the potential to affect analyses conducted in all fields of
scientific study, including healthcare, economics, and the social sciences.
Several approaches to unbiased inference in the presence of non-ignorable
missingness rely on the specification of the target distribution and its
missingness process as a probability distribution that factorizes with respect
to a directed acyclic graph. In this paper, we address the longstanding
question of the characterization of models that are identifiable within this
class of missing data distributions. We provide the first completeness result
in this field of study -- necessary and sufficient graphical conditions under
which, the full data distribution can be recovered from the observed data
distribution. We then simultaneously address issues that may arise due to the
presence of both missing data and unmeasured confounding, by extending these
graphical conditions and proofs of completeness, to settings where some
variables are not just missing, but completely unobserved.
| stat |
Self-Attention Generative Adversarial Networks | In this paper, we propose the Self-Attention Generative Adversarial Network
(SAGAN) which allows attention-driven, long-range dependency modeling for image
generation tasks. Traditional convolutional GANs generate high-resolution
details as a function of only spatially local points in lower-resolution
feature maps. In SAGAN, details can be generated using cues from all feature
locations. Moreover, the discriminator can check that highly detailed features
in distant portions of the image are consistent with each other. Furthermore,
recent work has shown that generator conditioning affects GAN performance.
Leveraging this insight, we apply spectral normalization to the GAN generator
and find that this improves training dynamics. The proposed SAGAN achieves the
state-of-the-art results, boosting the best published Inception score from 36.8
to 52.52 and reducing Frechet Inception distance from 27.62 to 18.65 on the
challenging ImageNet dataset. Visualization of the attention layers shows that
the generator leverages neighborhoods that correspond to object shapes rather
than local regions of fixed shape.
| stat |
Bernoulli Race Particle Filters | When the weights in a particle filter are not available analytically,
standard resampling methods cannot be employed. To circumvent this problem
state-of-the-art algorithms replace the true weights with non-negative unbiased
estimates. This algorithm is still valid but at the cost of higher variance of
the resulting filtering estimates in comparison to a particle filter using the
true weights. We propose here a novel algorithm that allows for resampling
according to the true intractable weights when only an unbiased estimator of
the weights is available. We demonstrate our algorithm on several examples.
| stat |
Gaussian Process Modulated Cox Processes under Linear Inequality
Constraints | Gaussian process (GP) modulated Cox processes are widely used to model point
patterns. Existing approaches require a mapping (link function) between the
unconstrained GP and the positive intensity function. This commonly yields
solutions that do not have a closed form or that are restricted to specific
covariance functions. We introduce a novel finite approximation of GP-modulated
Cox processes where positiveness conditions can be imposed directly on the GP,
with no restrictions on the covariance function. Our approach can also ensure
other types of inequality constraints (e.g. monotonicity, convexity), resulting
in more versatile models that can be used for other classes of point processes
(e.g. renewal processes). We demonstrate on both synthetic and real-world data
that our framework accurately infers the intensity functions. Where
monotonicity is a feature of the process, our ability to include this in the
inference improves results.
| stat |
Shapley Interpretation and Activation in Neural Networks | We propose a novel Shapley value approach to help address neural networks'
interpretability and "vanishing gradient" problems. Our method is based on an
accurate analytical approximation to the Shapley value of a neuron with ReLU
activation. This analytical approximation admits a linear propagation of
relevance across neural network layers, resulting in a simple, fast and
sensible interpretation of neural networks' decision making process.
We then derived a globally continuous and non-vanishing Shapley gradient,
which can replace the conventional gradient in training neural network layers
with ReLU activation, and leading to better training performance. We further
derived a Shapley Activation (SA) function, which is a close approximation to
ReLU but features the Shapley gradient. The SA is easy to implement in existing
machine learning frameworks. Numerical tests show that SA consistently
outperforms ReLU in training convergence, accuracy and stability.
| stat |
Regularized Variational Data Assimilation for Bias Treatment using the
Wasserstein Metric | This paper presents a new variational data assimilation (VDA) approach for
the formal treatment of bias in both model outputs and observations. This
approach relies on the Wasserstein metric stemming from the theory of optimal
mass transport to penalize the distance between the probability histograms of
the analysis state and an a priori reference dataset, which is likely to be
more uncertain but less biased than both model and observations. Unlike
previous bias-aware VDA approaches, the new Wasserstein metric VDA (WM-VDA)
dynamically treats systematic biases of unknown magnitude and sign in both
model and observations through assimilation of the reference data in the
probability domain and can fully recover the probability histogram of the
analysis state. The performance of WM-VDA is compared with the classic
three-dimensional VDA (3D-Var) scheme on first-order linear dynamics and the
chaotic Lorenz attractor. Under positive systematic biases in both model and
observations, we consistently demonstrate a significant reduction in the
forecast bias and unbiased root mean squared error.
| stat |
Online Predictive Optimization Framework for Stochastic
Demand-Responsive Transit Services | This study develops an online predictive optimization framework for
dynamically operating a transit service in an area of crowd movements. The
proposed framework integrates demand prediction and supply optimization to
periodically redesign the service routes based on recently observed demand. To
predict demand for the service, we use Quantile Regression to estimate the
marginal distribution of movement counts between each pair of serviced
locations. The framework then combines these marginals into a joint demand
distribution by constructing a Gaussian copula, which captures the structure of
correlation between the marginals. For supply optimization, we devise a linear
programming model, which simultaneously determines the route structure and the
service frequency according to the predicted demand. Importantly, our framework
both preserves the uncertainty structure of future demand and leverages this
for robust route optimization, while keeping both components decoupled. We
evaluate our framework using a real-world case study of autonomous mobility in
a university campus in Denmark. The results show that our framework often
obtains the ground truth optimal solution, and can outperform conventional
methods for route optimization, which do not leverage full predictive
distributions.
| stat |
Causal inference with Bayes rule | The concept of causality has a controversial history. The question of whether
it is possible to represent and address causal problems with probability
theory, or if fundamentally new mathematics such as the do-calculus is required
has been hotly debated, In this paper we demonstrate that, while it is critical
to explicitly model our assumptions on the impact of intervening in a system,
provided we do so, estimating causal effects can be done entirely within the
standard Bayesian paradigm. The invariance assumptions underlying causal
graphical models can be encoded in ordinary Probabilistic graphical models,
allowing causal estimation with Bayesian statistics, equivalent to the
do-calculus.
| stat |
Differentially Private Continual Learning | Catastrophic forgetting can be a significant problem for institutions that
must delete historic data for privacy reasons. For example, hospitals might not
be able to retain patient data permanently. But neural networks trained on
recent data alone will tend to forget lessons learned on old data. We present a
differentially private continual learning framework based on variational
inference. We estimate the likelihood of past data given the current model
using differentially private generative models of old datasets.
| stat |
Topological Uncertainty: Monitoring trained neural networks through
persistence of activation graphs | Although neural networks are capable of reaching astonishing performances on
a wide variety of contexts, properly training networks on complicated tasks
requires expertise and can be expensive from a computational perspective. In
industrial applications, data coming from an open-world setting might widely
differ from the benchmark datasets on which a network was trained. Being able
to monitor the presence of such variations without retraining the network is of
crucial importance. In this article, we develop a method to monitor trained
neural networks based on the topological properties of their activation graphs.
To each new observation, we assign a Topological Uncertainty, a score that aims
to assess the reliability of the predictions by investigating the whole network
instead of its final layer only, as typically done by practitioners. Our
approach entirely works at a post-training level and does not require any
assumption on the network architecture, optimization scheme, nor the use of
data augmentation or auxiliary datasets; and can be faithfully applied on a
large range of network architectures and data types. We showcase experimentally
the potential of Topological Uncertainty in the context of trained network
selection, Out-Of-Distribution detection, and shift-detection, both on
synthetic and real datasets of images and graphs.
| stat |
Efficient Primal-Dual Algorithms for Large-Scale Multiclass
Classification | We develop efficient algorithms to train $\ell_1$-regularized linear
classifiers with large dimensionality $d$ of the feature space, number of
classes $k$, and sample size $n$. Our focus is on a special class of losses
that includes, in particular, the multiclass hinge and logistic losses. Our
approach combines several ideas: (i) passing to the equivalent saddle-point
problem with a quasi-bilinear objective; (ii) applying stochastic mirror
descent with a proper choice of geometry which guarantees a favorable accuracy
bound; (iii) devising non-uniform sampling schemes to approximate the matrix
products. In particular, for the multiclass hinge loss we propose a
\textit{sublinear} algorithm with iterations performed in $O(d+n+k)$ arithmetic
operations.
| stat |
ROBustness In Network (robin): an R package for Comparison and
Validation of communities | In network analysis, many community detection algorithms have been developed,
however, their implementation leaves unaddressed the question of the
statistical validation of the results. Here we present robin(ROBustness In
Network), an R package to assess the robustness of the community structure of a
network found by one or more methods to give indications about their
reliability. The procedure initially detects if the community structure found
by a set of algorithms is statistically significant and then compares two
selected detection algorithms on the same graph to choose the one that better
fits the network of interest. We demonstrate the use of our package on the
American College Football benchmark dataset.
| stat |
A Multi-Stage Stochastic Programming Approach to Epidemic Resource
Allocation with Equity Considerations | Existing compartmental models in epidemiology are limited in terms of
optimizing the resource allocation to control an epidemic outbreak under
disease growth uncertainty. In this study, we address this core limitation by
presenting a multi-stage stochastic programming compartmental model, which
integrates the uncertain disease progression and resource allocation to control
an infectious disease outbreak. The proposed multi-stage stochastic program
involves various disease growth scenarios and optimizes the distribution of
treatment centers and resources while minimizing the total expected number of
new infections and funerals. We define two new equity metrics, namely infection
and capacity equity, and explicitly consider equity for allocating treatment
funds and facilities over multiple time stages. We also study the multi-stage
value of the stochastic solution (VSS), which demonstrates the superiority of
the proposed stochastic programming model over its deterministic counterpart.
We apply the proposed formulation to control the Ebola Virus Disease (EVD) in
Guinea, Sierra Leone, and Liberia of West Africa to determine the optimal and
fair resource-allocation strategies. Our model balances the proportion of
infections over all regions, even without including the infection equity or
prevalence equity constraints. Model results also show that allocating
treatment resources proportional to population is sub-optimal, and enforcing
such a resource allocation policy might adversely impact the total number of
infections and deaths, and thus resulting in a high cost that we have to pay
for the fairness. Our multi-stage stochastic epidemic-logistics model is
practical and can be adapted to control other infectious diseases in
meta-populations and dynamically evolving situations.
| stat |
A Survival Mediation Model with Bayesian Model Averaging | Determining the extent to which a patient is benefiting from cancer therapy
is challenging. Criteria for quantifying the extent of "tumor response"
observed within a few cycles of treatment have been established for various
types of solid as well as hematologic malignancies. These measures comprise the
primary endpoints of phase II trials. Regulatory approvals of new cancer
therapies, however, are usually contingent upon the demonstration of superior
overall survival with randomized evidence acquired with a phase III trial
comparing the novel therapy to an appropriate standard of care treatment. With
nearly two thirds of phase III oncology trials failing to achieve statistically
significant results, researchers continue to refine and propose new surrogate
endpoints. This article presents a Bayesian framework for studying
relationships among treatment, patient subgroups, tumor response and survival.
Combining classical components of mediation analysis with Bayesian model
averaging (BMA), the methodology is robust to model mis-specification among
various possible relationships among the observable entities. Posterior
inference is demonstrated via application to a randomized controlled phase III
trial in metastatic colorectal cancer. Moreover, the article details posterior
predictive distributions of survival and statistical metrics for quantifying
the extent of direct and indirect, or tumor response mediated, treatment
effects.
| stat |
Robust Causal Inference for Incremental Return on Ad Spend with
Randomized Paired Geo Experiments | Evaluating the incremental return on ad spend (iROAS) of a prospective online
marketing strategy---that is, the ratio of the strategy's causal effect on some
response metric of interest relative to its causal effect on the ad spend---has
become progressively more important as advertisers increasingly seek to better
understand the impact of their marketing decisions. Although randomized "geo
experiments" are frequently employed for this evaluation, obtaining reliable
estimates of the iROAS can be challenging as oftentimes only a small number of
highly heterogeneous units are used. In this paper, we formulate a novel
statistical framework for inferring the iROAS of online advertising in a
randomized paired geo experiment design which further motivates and provides
new insights into Rosenbaum's arguments on instrumental variables, and we
propose and develop a robust and distribution-free estimator "Trimmed Match"
which adaptively trims poorly matched pairs. Using numerical simulations and
real case studies, we show that Trimmed Match can be more efficient than some
alternatives, and we investigate the sensitivity of the estimator to some
violations of its assumptions. Consistency and asymptotic normality are also
established for a fixed trim rate. Technical proofs are provided in the
appendix.
| stat |
Semi-parametric Bayes Regression with Network Valued Covariates | There is an increasing recognition of the role of brain networks as
neuroimaging biomarkers in mental health and psychiatric studies. Our focus is
posttraumatic stress disorder (PTSD), where the brain network interacts with
environmental exposures in complex ways to drive the disease progression.
Existing linear models seeking to characterize the relation between the
clinical phenotype and the entire edge set in the brain network may be overly
simplistic and often involve inflated number of parameters leading to
computational burden and inaccurate estimation. In one of the first such
efforts, we develop a novel two stage Bayesian framework to find a
node-specific lower dimensional representation for the network using a latent
scale approach in the first stage, and then use a flexible Gaussian process
regression framework for prediction involving the latent scales and other
supplementary covariates in the second stage. The proposed approach relaxes
linearity assumptions, addresses the curse of dimensionality and is scalable to
high dimensional networks while maintaining interpretability at the node level
of the network. Extensive simulations and results from our motivating PTSD
application show a distinct advantage of the proposed approach over competing
linear and non-linear approaches in terms of prediction and coverage.
| stat |
Approximate Bayesian Bootstrap Procedures to Estimate Multilevel
Treatment Effects in Observational Studies with Application to Type 2
Diabetes Treatment Regimens | Randomized clinical trials are considered the gold standard for estimating
causal effects. Nevertheless, in studies that are aimed at examining adverse
effects of interventions, such trials are often impractical because of ethical
and financial considerations. In observational studies, matching on the
generalized propensity scores was proposed as a possible solution to estimate
the treatment effects of multiple interventions. However, the derivation of
point and interval estimates for these matching procedures can become complex
with non-continuous or censored outcomes. We propose a novel Approximate
Bayesian Bootstrap algorithm that result in statistically valid point and
interval estimates of the treatment effects with categorical outcomes. The
procedure relies on the estimated generalized propensity scores and multiply
imputes the unobserved potential outcomes for each unit. In addition, we
describe a corresponding interpretable sensitivity analysis to examine the
unconfoundedness assumption. We apply this approach to examines the
cardiovascular safety of common, real-world anti-diabetic treatment regimens
for Type 2 diabetes mellitus in a large observational database.
| stat |
Global Non-convex Optimization with Discretized Diffusions | An Euler discretization of the Langevin diffusion is known to converge to the
global minimizers of certain convex and non-convex optimization problems. We
show that this property holds for any suitably smooth diffusion and that
different diffusions are suitable for optimizing different classes of convex
and non-convex functions. This allows us to design diffusions suitable for
globally optimizing convex and non-convex functions not covered by the existing
Langevin theory. Our non-asymptotic analysis delivers computable optimization
and integration error bounds based on easily accessed properties of the
objective and chosen diffusion. Central to our approach are new explicit Stein
factor bounds on the solutions of Poisson equations. We complement these
results with improved optimization guarantees for targets other than the
standard Gibbs measure.
| stat |
A flexible adaptive lasso Cox frailty model based on the full likelihood | In this work a method to regularize Cox frailty models is proposed that
accommodates time-varying covariates and time-varying coefficients and is based
on the full instead of the partial likelihood. A particular advantage in this
framework is that the baseline hazard can be explicitly modeled in a smooth,
semi-parametric way, e.g. via P-splines. Regularization for variable selection
is performed via a lasso penalty and via group lasso for categorical variables
while a second penalty regularizes wiggliness of smooth estimates of
time-varying coefficients and the baseline hazard. Additionally, adaptive
weights are included to stabilize the estimation. The method is implemented in
R as coxlasso and will be compared to other packages for regularized Cox
regression. Existing packages, however, do not allow for the combination of
different effects that are accommodated in coxlasso.
| stat |
An Approximation Scheme for Multivariate Information based on Partial
Information Decomposition | We consider an approximation scheme for multivariate information assuming
that synergistic information only appearing in higher order joint distributions
is suppressed, which may hold in large classes of systems. Our approximation
scheme gives a practical way to evaluate information among random variables and
is expected to be applied to feature selection in machine learning. The
truncation order of our approximation scheme is given by the order of synergy.
In the classification of information, we use the partial information
decomposition of the original one. The resulting multivariate information is
expected to be reasonable if higher order synergy is suppressed in the system.
In addition, it is calculable in relatively easy way if the truncation order is
not so large. We also perform numerical experiments to check the validity of
our approximation scheme.
| stat |
Tensor Monte Carlo: particle methods for the GPU era | Multi-sample, importance-weighted variational autoencoders (IWAE) give
tighter bounds and more accurate uncertainty estimates than variational
autoencoders (VAE) trained with a standard single-sample objective. However,
IWAEs scale poorly: as the latent dimensionality grows, they require
exponentially many samples to retain the benefits of importance weighting.
While sequential Monte-Carlo (SMC) can address this problem, it is
prohibitively slow because the resampling step imposes sequential structure
which cannot be parallelised, and moreover, resampling is non-differentiable
which is problematic when learning approximate posteriors. To address these
issues, we developed tensor Monte-Carlo (TMC) which gives exponentially many
importance samples by separately drawing $K$ samples for each of the $n$ latent
variables, then averaging over all $K^n$ possible combinations. While the sum
over exponentially many terms might seem to be intractable, in many cases it
can be computed efficiently as a series of tensor inner-products. We show that
TMC is superior to IWAE on a generative model with multiple stochastic layers
trained on the MNIST handwritten digit database, and we show that TMC can be
combined with standard variance reduction techniques.
| stat |
Gradient Boosting Machine: A Survey | In this survey, we discuss several different types of gradient boosting
algorithms and illustrate their mathematical frameworks in detail: 1.
introduction of gradient boosting leads to 2. objective function optimization,
3. loss function estimations, and 4. model constructions. 5. application of
boosting in ranking.
| stat |
Distributed Bayesian clustering using finite mixture of mixtures | In many modern applications, there is interest in analyzing enormous data
sets that cannot be easily moved across computers or loaded into memory on a
single computer. In such settings, it is very common to be interested in
clustering. Existing distributed clustering algorithms are mostly distance or
density based without a likelihood specification, precluding the possibility of
formal statistical inference. Model-based clustering allows statistical
inference, yet research on distributed inference has emphasized nonparametric
Bayesian mixture models over finite mixture models. To fill this gap, we
introduce a nearly embarrassingly parallel algorithm for clustering under a
Bayesian overfitted finite mixture of Gaussian mixtures, which we term
distributed Bayesian clustering (DIB-C). DIB-C can flexibly accommodate data
sets with various shapes (e.g. skewed or multi-modal). With data randomly
partitioned and distributed, we first run Markov chain Monte Carlo in an
embarrassingly parallel manner to obtain local clustering draws and then refine
across workers for a final clustering estimate based on any loss function on
the space of partitions. DIB-C can also estimate cluster densities, quickly
classify new subjects and provide a posterior predictive distribution. Both
simulation studies and real data applications show superior performance of
DIB-C in terms of robustness and computational efficiency.
| stat |
Isomorphism Check for $2^n$ Factorial Designs with Randomization
Restrictions | Factorial designs with randomization restrictions are often used in
industrial experiments when a complete randomization of trials is impractical.
In the statistics literature, the analysis, construction and isomorphism of
factorial designs has been extensively investigated. Much of the work has been
on a case-by-case basis -- addressing completely randomized designs, randomized
block designs, split-plot designs, etc. separately. In this paper we take a
more unified approach, developing theoretical results and an efficient
relabeling strategy to both construct and check the isomorphism of multi-stage
factorial designs with randomization restrictions. The examples presented in
this paper particularly focus on split-lot designs.
| stat |
Deep Kernel Survival Analysis and Subject-Specific Survival Time
Prediction Intervals | Kernel survival analysis methods predict subject-specific survival curves and
times using information about which training subjects are most similar to a
test subject. These most similar training subjects could serve as forecast
evidence. How similar any two subjects are is given by the kernel function. In
this paper, we present the first neural network framework that learns which
kernel functions to use in kernel survival analysis. We also show how to use
kernel functions to construct prediction intervals of survival time estimates
that are statistically valid for individuals similar to a test subject. These
prediction intervals can use any kernel function, such as ones learned using
our neural kernel learning framework or using random survival forests. Our
experiments show that our neural kernel survival estimators are competitive
with a variety of existing survival analysis methods, and that our prediction
intervals can help compare different methods' uncertainties, even for
estimators that do not use kernels. In particular, these prediction interval
widths can be used as a new performance metric for survival analysis methods.
| stat |
A memory-free spatial additive mixed modeling for big spatial data | This study develops a spatial additive mixed modeling (AMM) approach
estimating spatial and non-spatial effects from large samples, such as millions
of observations. Although fast AMM approaches are already well-established,
they are restrictive in that they assume an known spatial dependence structure.
To overcome this limitation, this study develops a fast AMM with the estimation
of spatial structure in residuals and regression coefficients together with
non-spatial effects. We rely on a Moran coefficient-based approach to estimate
the spatial structure. The proposed approach pre-compresses large matrices
whose size grows with respect to the sample size N before the model estimation;
thus, the computational complexity for the estimation is independent of the
sample size. Furthermore, the pre-compression is done through a block-wise
procedure that makes the memory consumption independent of N. Eventually, the
spatial AMM is memory-free and fast even for millions of observations. The
developed approach is compared to alternatives through Monte Carlo simulation
experiments. The result confirms the accuracy and computational efficiency of
the developed approach. The developed approaches are implemented in an R
package spmoran.
| stat |
Minimum Volume Topic Modeling | We propose a new topic modeling procedure that takes advantage of the fact
that the Latent Dirichlet Allocation (LDA) log likelihood function is
asymptotically equivalent to the logarithm of the volume of the topic simplex.
This allows topic modeling to be reformulated as finding the probability
simplex that minimizes its volume and encloses the documents that are
represented as distributions over words. A convex relaxation of the minimum
volume topic model optimization is proposed, and it is shown that the relaxed
problem has the same global minimum as the original problem under the
separability assumption and the sufficiently scattered assumption introduced by
Arora et al. (2013) and Huang et al. (2016). A locally convergent alternating
direction method of multipliers (ADMM) approach is introduced for solving the
relaxed minimum volume problem. Numerical experiments illustrate the benefits
of our approach in terms of computation time and topic recovery performance.
| stat |
StructureBoost: Efficient Gradient Boosting for Structured Categorical
Variables | Gradient boosting methods based on Structured Categorical Decision Trees
(SCDT) have been demonstrated to outperform numerical and one-hot-encodings on
problems where the categorical variable has a known underlying structure.
However, the enumeration procedure in the SCDT is infeasible except for
categorical variables with low or moderate cardinality. We propose and
implement two methods to overcome the computational obstacles and efficiently
perform Gradient Boosting on complex structured categorical variables. The
resulting package, called StructureBoost, is shown to outperform established
packages such as CatBoost and LightGBM on problems with categorical predictors
that contain sophisticated structure. Moreover, we demonstrate that
StructureBoost can make accurate predictions on unseen categorical values due
to its knowledge of the underlying structure.
| stat |
Estimation and inference on high-dimensional individualized treatment
rule in observational data using split-and-pooled de-correlated score | With the increasing adoption of electronic health records, there is an
increasing interest in developing individualized treatment rules, which
recommend treatments according to patients' characteristics, from large
observational data. However, there is a lack of valid inference procedures for
such rules developed from this type of data in the presence of high-dimensional
covariates. In this work, we develop a penalized doubly robust method to
estimate the optimal individualized treatment rule from high-dimensional data.
We propose a split-and-pooled de-correlated score to construct hypothesis tests
and confidence intervals. Our proposal utilizes the data splitting to conquer
the slow convergence rate of nuisance parameter estimations, such as
non-parametric methods for outcome regression or propensity models. We
establish the limiting distributions of the split-and-pooled de-correlated
score test and the corresponding one-step estimator in high-dimensional
setting. Simulation and real data analysis are conducted to demonstrate the
superiority of the proposed method.
| stat |
Provably Strict Generalisation Benefit for Equivariant Models | It is widely believed that engineering a model to be invariant/equivariant
improves generalisation. Despite the growing popularity of this approach, a
precise characterisation of the generalisation benefit is lacking. By
considering the simplest case of linear models, this paper provides the first
provably non-zero improvement in generalisation for invariant/equivariant
models when the target distribution is invariant/equivariant with respect to a
compact group. Moreover, our work reveals an interesting relationship between
generalisation, the number of training examples and properties of the group
action. Our results rest on an observation of the structure of function spaces
under averaging operators which, along with its consequences for feature
averaging, may be of independent interest.
| stat |
Estimation of parameters of the Gumbel type-II distribution under AT-II
PHCS with an application of Covid-19 data | In this paper, we investigate the classical and Bayesian estimation of
unknown parameters of the Gumbel type-II distribution based on adaptive type-II
progressive hybrid censored sample (AT-II PHCS). The maximum likelihood
estimates (MLEs) and maximum product spacing estimates (MPSEs) are developed
and computed numerically using Newton-Raphson method. Bayesian approaches are
employed to estimate parameters under symmetric and asymmetric loss functions.
Bayesian estimates are not in explicit forms. Thus, Bayesian estimates are
obtained by using Markov chain Monte Carlo (MCMC) method along with the
Metropolis-Hastings (MH) algorithm. Based on the normality property of MLEs the
asymptotic confidence intervals are constructed. Also, bootstrap intervals and
highest posterior density (HPD) credible intervals are constructed. Further a
Monte Carlo simulation study is carried out. Finally, the data set based on the
death rate due to Covid-19 in India is analyzed for illustration of the
purpose.
| stat |
Model-based Differentially Private Data Synthesis and Statistical
Inference in Multiply Synthetic Differentially Private Data | We propose the approach of model-based differentially private synthesis
(modips) in the Bayesian framework for releasing individual-level
surrogate/synthetic datasets with privacy guarantees given the original data.
The modips technique integrates the concept of differential privacy into
model-based data synthesis. We introduce several variants for the general
modips approach and different procedures to obtaining privacy-preserving
posterior samples, a key step in modips. The uncertainty from the sanitization
and synthetic process in modips can be accounted for by releasing multiple
synthetic datasets and quantified via an inferential combination rule that is
proposed in this paper. We run empirical studies to examine the impacts of the
number of synthetic sets and the privacy budget allocation schemes on the
inference based on synthetic data.
| stat |
Constructed measures and causal inference: towards a new model of
measurement for psychosocial constructs | Psychosocial constructs can only be assessed indirectly, and measures are
typically formed by a combination of indicators that are thought to relate to
the construct. Reflective and formative measurement models offer different
conceptualizations of the relation between the indicators and what is sometimes
conceived of as a univariate latent variable supposed to correspond in some way
to the construct. It is argued that the empirical implications of reflective
and formative models will often be violated by data since the causally relevant
constituents will generally be multivariate, not univariate. These empirical
implications can be formally tested but factor analysis is not adequate to do
so. It is argued that formative models misconstrue the relationship between the
constructed measures and the underlying reality by which causal processes
operate, but that reflective models misconstrue the nature of the underlying
reality itself by typically presuming that the constituents of it that are
causally efficacious are unidimensional. The ensuing problems arising from
these misconstruals are discussed. A causal interpretation is proposed of
associations between constructed measures and various outcomes that is
applicable to both reflective and formative models and is applicable even if
the usual assumptions of these models are violated. An outline for a new model
of the process of measure construction is put forward. Discussion is given to
the practical implications of these observations and proposals for the
provision of definitions, the selection of items, item-by-item analyses, the
construction of measures, and the interpretation of the associations of these
measures with subsequent outcomes.
| stat |
Computationally efficient univariate filtering for massive data | The vast availability of large scale, massive and big data has increased the
computational cost of data analysis. One such case is the computational cost of
the univariate filtering which typically involves fitting many univariate
regression models and is essential for numerous variable selection algorithms
to reduce the number of predictor variables. The paper manifests how to
dramatically reduce that computational cost by employing the score test or the
simple Pearson correlation (or the t-test for binary responses). Extensive
Monte Carlo simulation studies will demonstrate their advantages and
disadvantages compared to the likelihood ratio test and examples with real data
will illustrate the performance of the score test and the log-likelihood ratio
test under realistic scenarios. Depending on the regression model used, the
score test is 30 - 60,000 times faster than the log-likelihood ratio test and
produces nearly the same results. Hence this paper strongly recommends to
substitute the log-likelihood ratio test with the score test when coping with
large scale data, massive data, big data, or even with data whose sample size
is in the order of a few tens of thousands or higher.
| stat |
The Autoregressive Structural Model for analyzing longitudinal health
data of an aging population in China | We seek to elucidate the impact of social activity, physical activity and
functional health status (factors) on depressive symptoms (outcome) in the
China Health and Retirement Longitudinal Study (CHARLS), a multi-year study of
aging involving 20,000 participants 45 years of age and older. Although a
variety of statistical methods are available for analyzing longitudinal data,
modeling the dynamics within a complex system remains a difficult
methodological challenge. We develop an Autoregressive Structural Model (ASM)
to examine these factors on depressive symptoms while accounting for temporal
dependence. The ASM builds on the structural equation model and also consists
of two components: a measurement model that connects observations to latent
factors, and a structural model that delineates the mechanism among latent
factors. Our ASM further incorporates autoregressive dependence into both
components for repeated measurements. The results from applying the ASM to the
CHARLS data indicate that social and physical activity independently and
consistently mitigated depressive symptoms over the course of five years, by
mediating through functional health status.
| stat |
Learning Deep Generative Models with Annealed Importance Sampling | Variational inference (VI) and Markov chain Monte Carlo (MCMC) are two main
approximate approaches for learning deep generative models by maximizing
marginal likelihood. In this paper, we propose using annealed importance
sampling for learning deep generative models. Our proposed approach bridges VI
with MCMC. It generalizes VI methods such as variational auto-encoders and
importance weighted auto-encoders (IWAE) and the MCMC method proposed in
(Hoffman, 2017). It also provides insights into why running multiple short MCMC
chains can help learning deep generative models. Through experiments, we show
that our approach yields better density models than IWAE and can effectively
trade computation for model accuracy without increasing memory cost.
| stat |
Rescuing neural spike train models from bad MLE | The standard approach to fitting an autoregressive spike train model is to
maximize the likelihood for one-step prediction. This maximum likelihood
estimation (MLE) often leads to models that perform poorly when generating
samples recursively for more than one time step. Moreover, the generated spike
trains can fail to capture important features of the data and even show
diverging firing rates. To alleviate this, we propose to directly minimize the
divergence between neural recorded and model generated spike trains using spike
train kernels. We develop a method that stochastically optimizes the maximum
mean discrepancy induced by the kernel. Experiments performed on both real and
synthetic neural data validate the proposed approach, showing that it leads to
well-behaving models. Using different combinations of spike train kernels, we
show that we can control the trade-off between different features which is
critical for dealing with model-mismatch.
| stat |
Deep Random Splines for Point Process Intensity Estimation of Neural
Population Data | Gaussian processes are the leading class of distributions on random
functions, but they suffer from well known issues including difficulty scaling
and inflexibility with respect to certain shape constraints (such as
nonnegativity). Here we propose Deep Random Splines, a flexible class of random
functions obtained by transforming Gaussian noise through a deep neural network
whose output are the parameters of a spline. Unlike Gaussian processes, Deep
Random Splines allow us to readily enforce shape constraints while inheriting
the richness and tractability of deep generative models. We also present an
observational model for point process data which uses Deep Random Splines to
model the intensity function of each point process and apply it to neural
population data to obtain a low-dimensional representation of spiking activity.
Inference is performed via a variational autoencoder that uses a novel
recurrent encoder architecture that can handle multiple point processes as
input. We use a newly collected dataset where a primate completes a pedaling
task, and observe better dimensionality reduction with our model than with
competing alternatives.
| stat |
An illustration of the risk of borrowing information via a shared
likelihood | A concrete, stylized example illustrates that inferences may be degraded,
rather than improved, by incorporating supplementary data via a joint
likelihood. In the example, the likelihood is assumed to be correctly
specified, as is the prior over the parameter of interest; all that is
necessary for the joint modeling approach to suffer is misspecification of the
prior over a nuisance parameter.
| stat |
Truncated, Censored, and Actuarial Payment-type Moments for Robust
Fitting of a Single-parameter Pareto Distribution | With some regularity conditions maximum likelihood estimators (MLEs) always
produce asymptotically optimal (in the sense of consistency, efficiency,
sufficiency, and unbiasedness) estimators. But in general, the MLEs lead to
non-robust statistical inference, for example, pricing models and risk
measures. Actuarial claim severity is continuous, right-skewed, and frequently
heavy-tailed. The data sets that such models are usually fitted to contain
outliers that are difficult to identify and separate from genuine data.
Moreover, due to commonly used actuarial "loss control strategies" in financial
and insurance industries, the random variables we observe and wish to model are
affected by truncation (due to deductibles), censoring (due to policy limits),
scaling (due to coinsurance proportions) and other transformations. To
alleviate the lack of robustness of MLE-based inference in risk modeling, here
in this paper, we propose and develop a new method of estimation - method of
truncated moments (MTuM) and generalize it for different scenarios of loss
control mechanism. Various asymptotic properties of those estimates are
established by using central limit theory. New connections between different
estimators are found. A comparative study of newly-designed methods with the
corresponding MLEs is performed. Detail investigation has been done for a
single parameter Pareto loss model including a simulation study.
| stat |
Investigating the Robustness of Artificial Intelligent Algorithms with
Mixture Experiments | Artificial intelligent (AI) algorithms, such as deep learning and XGboost,
are used in numerous applications including computer vision, autonomous
driving, and medical diagnostics. The robustness of these AI algorithms is of
great interest as inaccurate prediction could result in safety concerns and
limit the adoption of AI systems. In this paper, we propose a framework based
on design of experiments to systematically investigate the robustness of AI
classification algorithms. A robust classification algorithm is expected to
have high accuracy and low variability under different application scenarios.
The robustness can be affected by a wide range of factors such as the imbalance
of class labels in the training dataset, the chosen prediction algorithm, the
chosen dataset of the application, and a change of distribution in the training
and test datasets. To investigate the robustness of AI classification
algorithms, we conduct a comprehensive set of mixture experiments to collect
prediction performance results. Then statistical analyses are conducted to
understand how various factors affect the robustness of AI classification
algorithms. We summarize our findings and provide suggestions to practitioners
in AI applications.
| stat |
Design-adherent estimators for network surveys | Network surveys of key populations at risk for HIV are an essential part of
the effort to understand how the epidemic spreads and how it can be prevented.
Estimation of population values from the sample data has been probematical,
however, because the link-tracing of the network surveys includes different
people in the sample with unequal probabilities, and these inclusion
probabilities have to be estimated accurately to avoid large biases in survey
estimates. A new approach to estimation is introduced here, based on resampling
the sample network many times using a design that adheres to main features of
the design used in the field. These features include network link tracing,
branching, and without-replacement sampling. The frequency that a person is
included in the resamples is used to estimate the inclusion probability for
each person in the original sample, and these estimates of inclusion
probabilities are used in an unequal-probability estimator. In simulations
using a population of drug users, sex workers, and their partners for which the
actual values of population characteristics are known, the design-adherent
estimation approach increases the accuracy of estimates of population
quantities, largely by eliminating most of the biases.
| stat |
Efficient Estimation of Mixture Cure Frailty Model for Clustered Current
Status Data | Current status data abounds in the field of epidemiology and public health,
where the only observable data for a subject is the random inspection time and
the event status at inspection. Motivated by such a current status data from a
periodontal study where data are inherently clustered, we propose a unified
methodology to analyze such complex data. We allow the time-to-event to follow
the semiparametric GOR model with a cure fraction, and develop a unified
estimation scheme powered by the EM algorithm. The within-subject correlation
is accounted for by a random (frailty) effect, and the non-parametric component
of the GOR model is approximated via penalized splines, with a set of knot
points that increases with the sample size. Proposed methodology is accompanied
by a rigorous asymptotic theory, and the related semiparametric efficiency. The
finite sample performance of our model parameters are assessed via simulation
studies. Furthermore, the proposed methodology is illustrated via application
to the oral health data, accompanied by diagnostic checks to identify
influential observations. An easy to use R package CRFCSD is also available for
implementation.
| stat |
Bayesian analysis of Turkish Income and Living Conditions data, using
clustered longitudinal ordinal modelling with Bridge distributed
random-effects | This paper is motivated by the panel surveys, called Statistics on Income and
Living Conditions (SILC), conducted annually on (randomly selected)
country-representative households to monitor EU 2020 aims on poverty reduction.
We particularly consider the surveys conducted in Turkey, within the scope of
integration to the EU, between 2010 and 2013. Our main interests are on health
aspects of economic and living conditions. The outcome is {\it self-reported
health} that is clustered longitudinal ordinal, since repeated measures of it
are nested within individuals and individuals are nested within families.
Economic and living conditions were measured through a number of individual-
and family-level explanatory variables. The questions of interest are on the
marginal relationships between the outcome and covariates that are addressed
using a polytomous logistic regression with Bridge distributed random-effects.
This choice of distribution allows one to {\it directly} obtain marginal
inferences in the presence of random-effects. Widely used Normal distribution
is also considered as the random-effects distribution. Samples from the joint
posterior density of parameters and random-effects are drawn using Markov Chain
Monte Carlo. Interesting findings from public health point of view are that
differences were found between sub-groups of employment status, income level
and panel year in terms of odds of reporting better health.
| stat |
Estimating the number of SARS-CoV-2 infections and the impact of social
distancing in the United States | Understanding the number of individuals who have been infected with the novel
coronavirus SARS-CoV-2, and the extent to which social distancing policies have
been effective at limiting its spread, are critical for effective policy going
forward. Here we present estimates of the extent to which confirmed cases in
the United States undercount the true number of infections, and analyze how
effective social distancing measures have been at mitigating or suppressing the
virus. Our analysis uses a Bayesian model of COVID-19 fatalities with a
likelihood based on an underlying differential equation model of the epidemic.
We provide analysis for four states with significant epidemics: California,
Florida, New York, and Washington. Our short-term forecasts suggest that these
states may be following somewhat different trajectories for growth of the
number of cases and fatalities.
| stat |
Tracking change-points in multivariate extremes | In this paper we devise a statistical method for tracking and modeling
change-points on the dependence structure of multivariate extremes. The methods
are motivated by and illustrated on a case study on crypto-assets.
| stat |
Shallow Neural Hawkes: Non-parametric kernel estimation for Hawkes
processes | Multi-dimensional Hawkes process (MHP) is a class of self and mutually
exciting point processes that find wide range of applications -- from
prediction of earthquakes to modelling of order books in high frequency
trading. This paper makes two major contributions, we first find an unbiased
estimator for the log-likelihood estimator of the Hawkes process to enable
efficient use of the stochastic gradient descent method for maximum likelihood
estimation. The second contribution is, we propose a specific single hidden
layered neural network for the non-parametric estimation of the underlying
kernels of the MHP. We evaluate the proposed model on both synthetic and real
datasets, and find the method has comparable or better performance than
existing estimation methods. The use of shallow neural network ensures that we
do not compromise on the interpretability of the Hawkes model, while at the
same time have the flexibility to estimate any non-standard Hawkes excitation
kernel.
| stat |
Bayesian Double Feature Allocation for Phenotyping with Electronic
Health Records | We propose a categorical matrix factorization method to infer latent diseases
from electronic health records (EHR) data in an unsupervised manner. A latent
disease is defined as an unknown biological aberration that causes a set of
common symptoms for a group of patients. The proposed approach is based on a
novel double feature allocation model which simultaneously allocates features
to the rows and the columns of a categorical matrix. Using a Bayesian approach,
available prior information on known diseases greatly improves identifiability
and interpretability of latent diseases. This includes known diagnoses for
patients and known association of diseases with symptoms. We validate the
proposed approach by simulation studies including mis-specified models and
comparison with sparse latent factor models. In the application to Chinese EHR
data, we find interesting results, some of which agree with related clinical
and medical knowledge.
| stat |
Temporal models for demographic and global health outcomes in multiple
populations: Introducing a new framework to review and standardize
documentation of model assumptions and facilitate model comparison | There is growing interest in producing estimates of demographic and global
health indicators in populations with limited data. Statistical models are
needed to combine data from multiple data sources into estimates and
projections with uncertainty. Diverse modeling approaches have been applied to
this problem, making comparisons between models difficult. We propose a model
class, Temporal Models for Multiple Populations (TMMPs), to facilitate
documentation of model assumptions in a standardized way and comparison across
models. The class distinguishes between latent trends and the observed data,
which may be noisy or exhibit systematic biases. We provide general
formulations of the process model, which describes the latent trend of the
indicator of interest. We show how existing models for a variety of indicators
can be written as TMMPs and how the TMMP-based description can be used to
compare and contrast model assumptions. We end with a discussion of outstanding
questions and future directions.
| stat |
Discovering Interactions Using Covariate Informed Random Partition
Models | Combination chemotherapy treatment regimens created for patients diagnosed
with childhood acute lymphoblastic leukemia have had great success in improving
cure rates. Unfortunately, patients prescribed these types of treatment
regimens have displayed susceptibility to the onset of osteonecrosis. Some have
suggested that this is due to pharmacokinetic interaction between two agents in
the treatment regimen (asparaginase and dexamethasone) and other physiological
variables. Determining which physiological variables to consider when searching
for interactions in scenarios like these, minus a priori guidance, has proved
to be a challenging problem, particularly if interactions influence the
response distribution in ways beyond shifts in expectation or dispersion only.
In this paper we propose an exploratory technique that is able to discover
associations between covariates and responses in a very general way. The
procedure connects covariates to responses very flexibly through dependent
random partition prior distributions, and then employs machine learning
techniques to highlight potential associations found in each cluster. We
provide a simulation study to show utility and apply the method to data
produced from a study dedicated to learning which physiological predictors
influence severity of osteonecrosis multiplicatively.
| stat |
Accelerating Metropolis-within-Gibbs sampler with localized computations
of differential equations | Inverse problem is ubiquitous in science and engineering, and Bayesian
methodologies are often used to infer the underlying parameters. For high
dimensional temporal-spatial models, classical Markov chain Monte Carlo (MCMC)
methods are often slow to converge, and it is necessary to apply
Metropolis-within-Gibbs (MwG) sampling on parameter blocks. However, the
computation cost of each MwG iteration is typically $O(n^2)$, where $n$ is the
model dimension. This can be too expensive in practice. This paper introduces a
new reduced computation method to bring down the computation cost to $O(n)$,
for the inverse initial value problem of a stochastic differential equation
(SDE) with local interactions. The key observation is that each MwG proposal is
only different from the original iterate at one parameter block, and this
difference will only propagate within a local domain in the SDE computations.
Therefore we can approximate the global SDE computation with a surrogate
updated only within the local domain for reduced computation cost. Both
theoretically and numerically, we show that the approximation errors can be
controlled by the local domain size. We discuss how to implement the local
computation scheme using Euler--Maruyama and 4th order Runge--Kutta methods. We
numerically demonstrate the performance of the proposed method with the Lorenz
96 model and a linear stochastic flow model.
| stat |
On expectile-assisted inverse regression estimation for sufficient
dimension reduction | Moment-based sufficient dimension reduction methods such as sliced inverse
regression may not work well in the presence of heteroscedasticity. We propose
to first estimate the expectiles through kernel expectile regression, and then
carry out dimension reduction based on random projections of the regression
expectiles. Several popular inverse regression methods in the literature are
extended under this general framework. The proposed expectile-assisted methods
outperform existing moment-based dimension reduction methods in both numerical
studies and an analysis of the Big Mac data.
| stat |
Root-finding Approaches for Computing Conformal Prediction Set | Conformal prediction constructs a confidence region for an unobserved
response of a feature vector based on previous identically distributed and
exchangeable observations of responses and features. It has a coverage
guarantee at any nominal level without additional assumptions on their
distribution. However, it requires a refitting procedure for all replacement
candidates of the target response. In regression settings, this corresponds to
an infinite number of model fit. Apart from relatively simple estimators that
can be written as pieces of linear function of the response, efficiently
computing such sets is difficult and is still considered as an open problem. We
exploit the fact that, \emph{often}, conformal prediction sets are intervals
whose boundaries can be efficiently approximated by classical root-finding
software. We investigate how this approach can overcome many limitations of
formerly used strategies and achieves calculations that have been unattainable
so far. We discuss its complexity as well as its drawbacks and evaluate its
efficiency through numerical experiments.
| stat |
On models for the estimation of the excess mortality hazard in case of
insufficiently stratified life tables | In cancer epidemiology using population-based data, regression models for the
excess mortality hazard is a useful method to estimate cancer survival and to
describe the association between prognosis factors and excess mortality. This
method requires expected mortality rates from general population life tables:
each cancer patient is assigned an expected (background) mortality rate
obtained from the life tables, typically at least according to their age and
sex, from the population they belong to. However, those life tables may be
insufficiently stratified, as some characteristics such as deprivation,
ethnicity, and comorbidities, are not available in the life tables for a number
of countries. This may affect the background mortality rate allocated to each
patient, and it has been shown that not including relevant information for
assigning an expected mortality rate to each patient induces a bias in the
estimation of the regression parameters of the excess hazard model. We propose
two parametric corrections in excess hazard regression models, including a
single-parameter or a random effect (frailty), to account for possible
mismatches in the life table and thus misspecification of the background
mortality rate. In an extensive simulation study, the good statistical
performance of the proposed approach is demonstrated, and we illustrate their
use on real population-based data of lung cancer patients. We present
conditions and limitations of these methods, and provide some recommendations
for their use in practice.
| stat |
A weighted random survival forest | A weighted random survival forest is presented in the paper. It can be
regarded as a modification of the random forest improving its performance. The
main idea underlying the proposed model is to replace the standard procedure of
averaging used for estimation of the random survival forest hazard function by
weighted avaraging where the weights are assigned to every tree and can be
veiwed as training paremeters which are computed in an optimal way by solving a
standard quadratic optimization problem maximizing Harrell's C-index. Numerical
examples with real data illustrate the outperformance of the proposed model in
comparison with the original random survival forest.
| stat |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.