Update Space (evaluate main: 828c6327)

README.md

@@ -1,12 +1,126 @@
 ---
-title: Mae
-emoji: 📈
-colorFrom: gray
-colorTo: pink
+title: MAE
+emoji: 🤗 
+colorFrom: blue
+colorTo: red
 sdk: gradio
 sdk_version: 3.0.2
 app_file: app.py
 pinned: false
+tags:
+- evaluate
+- metric
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces#reference
+# Metric Card for MAE
+
+
+## Metric Description
+
+Mean Absolute Error (MAE) is the mean of the magnitude of difference between the predicted and actual numeric values:
+![image](https://user-images.githubusercontent.com/14205986/165824243-e1078dfd-489d-456c-a0da-cbaa28726220.png)
+
+
+## How to Use
+
+At minimum, this metric requires predictions and references as inputs.
+
+```python
+>>> mae_metric = evaluate.load("mae")
+>>> predictions = [2.5, 0.0, 2, 8]
+>>> references = [3, -0.5, 2, 7]
+>>> results = mae_metric.compute(predictions=predictions, references=references)
+```
+
+### Inputs
+
+Mandatory inputs: 
+- `predictions`: numeric array-like of shape (`n_samples,`) or (`n_samples`, `n_outputs`), representing the estimated target values.
+- `references`: numeric array-like of shape (`n_samples,`) or (`n_samples`, `n_outputs`), representing the ground truth (correct) target values.
+
+Optional arguments:
+- `sample_weight`: numeric array-like of shape (`n_samples,`) representing sample weights. The default is `None`.
+- `multioutput`: `raw_values`, `uniform_average` or numeric array-like of shape (`n_outputs,`), which defines the aggregation of multiple output values. The default value is `uniform_average`.
+  - `raw_values` returns a full set of errors in case of multioutput input.
+  - `uniform_average` means that the errors of all outputs are averaged with uniform weight. 
+  - the array-like value defines weights used to average errors.
+
+### Output Values
+This metric outputs a dictionary, containing the mean absolute error score, which is of type:
+- `float`: if multioutput is `uniform_average` or an ndarray of weights, then the weighted average of all output errors is returned.
+- numeric array-like of shape (`n_outputs,`): if multioutput is `raw_values`, then the score is returned for each output separately. 
+
+Each MAE `float` value ranges from `0.0` to `1.0`, with the best value being 0.0.
+
+Output Example(s):
+```python
+{'mae': 0.5}
+```
+
+If `multioutput="raw_values"`:
+```python
+{'mae': array([0.5, 1. ])}
+```
+
+#### Values from Popular Papers
+
+
+### Examples
+
+Example with the `uniform_average` config:
+```python
+>>> mae_metric = evaluate.load("mae")
+>>> predictions = [2.5, 0.0, 2, 8]
+>>> references = [3, -0.5, 2, 7]
+>>> results = mae_metric.compute(predictions=predictions, references=references)
+>>> print(results)
+{'mae': 0.5}
+```
+
+Example with multi-dimensional lists, and the `raw_values` config:
+```python
+>>> mae_metric = evaluate.load("mae", "multilist")
+>>> predictions = [[0.5, 1], [-1, 1], [7, -6]]
+>>> references = [[0, 2], [-1, 2], [8, -5]]
+>>> results = mae_metric.compute(predictions=predictions, references=references)
+>>> print(results)
+{'mae': 0.75}
+>>> results = mae_metric.compute(predictions=predictions, references=references, multioutput='raw_values')
+>>> print(results)
+{'mae': array([0.5, 1. ])}
+```
+
+## Limitations and Bias
+One limitation of MAE is that the relative size of the error is not always obvious, meaning that it can be difficult to tell a big error from a smaller one -- metrics such as Mean Absolute Percentage Error (MAPE) have been proposed to calculate MAE in percentage terms.
+
+Also, since it calculates the mean, MAE may underestimate the impact of big, but infrequent, errors -- metrics such as the Root Mean Square Error (RMSE) compensate for this by taking the root of error values. 
+
+## Citation(s)
+```bibtex
+@article{scikit-learn,
+  title={Scikit-learn: Machine Learning in {P}ython},
+  author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.
+         and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.
+         and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and
+         Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},
+  journal={Journal of Machine Learning Research},
+  volume={12},
+  pages={2825--2830},
+  year={2011}
+}
+```
+
+```bibtex
+@article{willmott2005advantages,
+  title={Advantages of the mean absolute error (MAE) over the root mean square error (RMSE) in assessing average model performance},
+  author={Willmott, Cort J and Matsuura, Kenji},
+  journal={Climate research},
+  volume={30},
+  number={1},
+  pages={79--82},
+  year={2005}
+}
+```
+
+## Further References
+- [Mean Absolute Error - Wikipedia](https://en.wikipedia.org/wiki/Mean_absolute_error)

app.py

@@ -0,0 +1,6 @@
+import evaluate
+from evaluate.utils import launch_gradio_widget
+
+
+module = evaluate.load("mae")
+launch_gradio_widget(module)

mae.py

@@ -0,0 +1,113 @@
+# Copyright 2022 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""MAE - Mean Absolute Error Metric"""
+
+import datasets
+from sklearn.metrics import mean_absolute_error
+
+import evaluate
+
+
+_CITATION = """\
+@article{scikit-learn,
+ title={Scikit-learn: Machine Learning in {P}ython},
+ author={Pedregosa, F. and Varoquaux, G. and Gramfort, A. and Michel, V.
+         and Thirion, B. and Grisel, O. and Blondel, M. and Prettenhofer, P.
+         and Weiss, R. and Dubourg, V. and Vanderplas, J. and Passos, A. and
+         Cournapeau, D. and Brucher, M. and Perrot, M. and Duchesnay, E.},
+ journal={Journal of Machine Learning Research},
+ volume={12},
+ pages={2825--2830},
+ year={2011}
+}
+"""
+
+_DESCRIPTION = """\
+Mean Absolute Error (MAE) is the mean of the magnitude of difference between the predicted and actual
+values.
+"""
+
+
+_KWARGS_DESCRIPTION = """
+Args:
+    predictions: array-like of shape (n_samples,) or (n_samples, n_outputs)
+        Estimated target values.
+    references: array-like of shape (n_samples,) or (n_samples, n_outputs)
+        Ground truth (correct) target values.
+    sample_weight: array-like of shape (n_samples,), default=None
+        Sample weights.
+    multioutput: {"raw_values", "uniform_average"} or array-like of shape (n_outputs,), default="uniform_average"
+        Defines aggregating of multiple output values. Array-like value defines weights used to average errors.
+
+                 "raw_values" : Returns a full set of errors in case of multioutput input.
+
+                 "uniform_average" : Errors of all outputs are averaged with uniform weight.
+
+Returns:
+    mae : mean absolute error.
+        If multioutput is "raw_values", then mean absolute error is returned for each output separately. If multioutput is "uniform_average" or an ndarray of weights, then the weighted average of all output errors is returned.
+        MAE output is non-negative floating point. The best value is 0.0.
+Examples:
+
+    >>> mae_metric = evaluate.load("mae")
+    >>> predictions = [2.5, 0.0, 2, 8]
+    >>> references = [3, -0.5, 2, 7]
+    >>> results = mae_metric.compute(predictions=predictions, references=references)
+    >>> print(results)
+    {'mae': 0.5}
+
+    If you're using multi-dimensional lists, then set the config as follows :
+
+    >>> mae_metric = evaluate.load("mae", "multilist")
+    >>> predictions = [[0.5, 1], [-1, 1], [7, -6]]
+    >>> references = [[0, 2], [-1, 2], [8, -5]]
+    >>> results = mae_metric.compute(predictions=predictions, references=references)
+    >>> print(results)
+    {'mae': 0.75}
+    >>> results = mae_metric.compute(predictions=predictions, references=references, multioutput='raw_values')
+    >>> print(results)
+    {'mae': array([0.5, 1. ])}
+"""
+
+
+@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
+class Mae(evaluate.EvaluationModule):
+    def _info(self):
+        return evaluate.EvaluationModuleInfo(
+            description=_DESCRIPTION,
+            citation=_CITATION,
+            inputs_description=_KWARGS_DESCRIPTION,
+            features=datasets.Features(self._get_feature_types()),
+            reference_urls=[
+                "https://scikit-learn.org/stable/modules/generated/sklearn.metrics.mean_absolute_error.html"
+            ],
+        )
+
+    def _get_feature_types(self):
+        if self.config_name == "multilist":
+            return {
+                "predictions": datasets.Sequence(datasets.Value("float")),
+                "references": datasets.Sequence(datasets.Value("float")),
+            }
+        else:
+            return {
+                "predictions": datasets.Value("float"),
+                "references": datasets.Value("float"),
+            }
+
+    def _compute(self, predictions, references, sample_weight=None, multioutput="uniform_average"):
+
+        mae_score = mean_absolute_error(references, predictions, sample_weight=sample_weight, multioutput=multioutput)
+
+        return {"mae": mae_score}

requirements.txt

@@ -0,0 +1,4 @@
+# TODO: fix github to release
+git+https://github.com/huggingface/evaluate.git@b6e6ed7f3e6844b297bff1b43a1b4be0709b9671
+datasets~=2.0
+sklearn