Merge branch 'main' into gaugup/AddLinkCheck

.flake8

@@ -0,0 +1,2 @@
+[flake8]
+max-line-length = 127

.github/workflows/notebook-linting.yml

@@ -0,0 +1,31 @@
+# This workflow will lint jupyter notebooks with flake8-nb.
+# For more information see: https://help.github.com/actions/language-and-framework-guides/using-python-with-github-actions
+
+name: Notebook linting
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+  schedule:
+    - cron:  '30 5 * * *'
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python 3.11
+      uses: actions/setup-python@v4
+      with:
+        python-version: "3.11"
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install flake8-nb==0.4.0
+    - name: Lint notebooks with flake8_nb
+      run: |
+        # stop the build if there are flake8 errors in notebooks
+        flake8_nb docs/source/notebooks/ --statistics --max-line-length=127

.github/workflows/notebook-tests.yml

@@ -0,0 +1,46 @@
+# This workflow will install Python dependencies, run tests and lint with a variety of Python versions
+# For more information see: https://help.github.com/actions/language-and-framework-guides/using-python-with-github-actions
+
+name: Notebook tests
+
+on:
+  push:
+    branches: [ main ]
+  pull_request:
+    branches: [ main ]
+  schedule:
+    - cron:  '30 5 * * *'
+
+jobs:
+  build:
+
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11"]
+        os: [ubuntu-latest, macos-latest]
+        exclude:
+          - os: macos-latest
+            python-version: "3.7"
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python ${{ matrix.python-version }} ${{ matrix.os }}
+      uses: actions/setup-python@v4
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Upgrade pip
+      run: |
+        python -m pip install --upgrade pip
+    - name: Install core dependencies
+      run: |
+        pip install -r requirements.txt
+    - name: Install deep learning dependencies
+      run: |
+        pip install -r requirements-deeplearning.txt
+    - name: Install test dependencies
+      run: |
+        pip install -r requirements-test.txt
+    - name: Test with pytest
+      run: |
+        # pytest
+        pytest tests/ -m "notebook_tests" --durations=10 --doctest-modules --junitxml=junit/test-results.xml --cov=dice_ml --cov-report=xml --cov-report=html

.github/workflows/python-linting.yml

@@ -1,4 +1,4 @@
-# This workflow will lint python code with flake8 and flake8-nb.
+# This workflow will lint python code with flake8.
 # For more information see: https://help.github.com/actions/language-and-framework-guides/using-python-with-github-actions
 
 name: Python linting
@@ -17,10 +17,10 @@ jobs:
 
     steps:
     - uses: actions/checkout@v3
-    - name: Set up Python 3.7
+    - name: Set up Python 3.11
       uses: actions/setup-python@v4
       with:
-        python-version: 3.7
+        python-version: "3.11"
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
@@ -37,7 +37,4 @@ jobs:
         # Check for cyclometric complexity for specific files where this metric has been 
         # reduced to ten and below
         flake8 dice_ml/data_interfaces/ --count --max-complexity=10 --max-line-length=127
-    - name: Lint notebooks with flake8_nb
-      run: |
-        # stop the build if there are flake8 errors in notebooks
-        flake8_nb docs/source/notebooks/ --statistics --max-line-length=127
+

.github/workflows/pythonpackage.yml → .github/workflows/python-package.yml

@@ -17,8 +17,11 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: [3.7, 3.8, 3.9, "3.10"]
-        os: [ubuntu-latest, macos-latest]
+        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11"]
+        os: [ubuntu-latest, macos-latest, windows-latest]
+        exclude:
+          - os: macos-latest
+            python-version: "3.7"
     steps:
     - uses: actions/checkout@v3
     - name: Set up Python ${{ matrix.python-version }} ${{ matrix.os }}
@@ -40,14 +43,14 @@ jobs:
     - name: Test with pytest
       run: |
         # pytest
-        pytest tests/ --durations=10 --doctest-modules --junitxml=junit/test-results.xml --cov=dice_ml --cov-report=xml --cov-report=html
+        pytest tests/ -m "not notebook_tests" --durations=10 --doctest-modules --junitxml=junit/test-results.xml --cov=dice_ml --cov-report=xml --cov-report=html
     - name: Publish Unit Test Results
       uses: EnricoMi/publish-unit-test-result-action/composite@v1
       if: ${{ (matrix.python-version == '3.8') && (matrix.os == 'ubuntu-latest') }}
       with:
         files: junit/test-results.xml
     - name: Upload coverage to Codecov
-      uses: codecov/codecov-action@v2
+      uses: codecov/codecov-action@v3
       if: ${{ (matrix.python-version == '3.8') && (matrix.os == 'ubuntu-latest') }}
       with:
         token: ${{ secrets.CODECOV_TOKEN }}

README.rst

@@ -1,11 +1,17 @@
-|BuildStatus|_ |PyPiVersion|_ |PythonSupport|_ |CondaVersion|_
+|BuildStatusTests|_ |BuildStatusNotebooks|_ |PyPiVersion|_ |PythonSupport|_ |Downloads|_ |CondaVersion|_
 
-.. |BuildStatus| image:: https://github.com/interpretml/dice/workflows/Python%20package/badge.svg
-.. _BuildStatus: https://github.com/interpretml/dice/actions?query=workflow%3A%22Python+package%22
+.. |BuildStatusTests| image:: https://github.com/interpretml/DiCE/actions/workflows/python-package.yml/badge.svg?branch=main
+.. _BuildStatusTests: https://github.com/interpretml/DiCE/actions/workflows/python-package.yml?query=workflow%3A%22Python+package%22
+
+.. |BuildStatusNotebooks| image:: https://github.com/interpretml/DiCE/actions/workflows/notebook-tests.yml/badge.svg?branch=main
+.. _BuildStatusNotebooks: https://github.com/interpretml/DiCE/actions/workflows/notebook-tests.yml?query=workflow%3A%22Notebook+tests%22
 
 .. |PyPiVersion| image:: https://img.shields.io/pypi/v/dice-ml
 .. _PyPiVersion: https://pypi.org/project/dice-ml/
 
+.. |Downloads| image:: https://static.pepy.tech/personalized-badge/dice-ml?period=total&units=international_system&left_color=grey&right_color=orange&left_text=Downloads
+.. _Downloads: https://pepy.tech/project/dice-ml
+
 .. |PythonSupport| image:: https://img.shields.io/pypi/pyversions/dice-ml
 .. _PythonSupport: https://pypi.org/project/dice-ml/
 
@@ -35,7 +41,7 @@ Diverse Counterfactual Explanations (DiCE) for ML
 Explanations are critical for machine learning, especially as machine learning-based systems are being used to inform decisions in societally critical domains such as finance, healthcare, education, and criminal justice.
 However, most explanation methods depend on an approximation of the ML model to
 create an interpretable explanation. For example,
-consider a person who applied for a loan and was rejected by the loan distribution algorithm of a financial company. Typically, the company may provide an explanation on why the loan was rejected, for example, due to "poor credit history". However, such an explanation does not help the person decide *what they do should next* to improve their chances of being approved in the future. Critically, the most important feature may not be enough to flip the decision of the algorithm, and in practice, may not even be changeable such as gender and race.
+consider a person who applied for a loan and was rejected by the loan distribution algorithm of a financial company. Typically, the company may provide an explanation on why the loan was rejected, for example, due to "poor credit history". However, such an explanation does not help the person decide *what they should do next* to improve their chances of being approved in the future. Critically, the most important feature may not be enough to flip the decision of the algorithm, and in practice, may not even be changeable such as gender and race.
 
 
 DiCE implements `counterfactual (CF) explanations <https://arxiv.org/abs/1711.00399>`_  that provide this information by showing feature-perturbed versions of the same person who would have received the loan, e.g., ``you would have received the loan if your income was higher by $10,000``. In other words, it provides "what-if" explanations for model output and can be a useful complement to other explanation methods, both for end-users and model developers.

dice_ml/counterfactual_explanations.py

@@ -2,11 +2,11 @@
 import os
 
 import jsonschema
+from raiutils.exceptions import UserConfigValidationException
 
 from dice_ml.constants import _SchemaVersions
 from dice_ml.diverse_counterfactuals import (CounterfactualExamples,
                                              _DiverseCFV2SchemaConstants)
-from dice_ml.utils.exception import UserConfigValidationException
 
 
 class _CommonSchemaConstants:

dice_ml/data_interfaces/base_data_interface.py

@@ -2,8 +2,9 @@
 
 from abc import ABC, abstractmethod
 
-from dice_ml.utils.exception import (SystemException,
-                                     UserConfigValidationException)
+from raiutils.exceptions import UserConfigValidationException
+
+from dice_ml.utils.exception import SystemException
 
 
 class _BaseData(ABC):

dice_ml/data_interfaces/private_data_interface.py

@@ -379,7 +379,7 @@ def get_ohe_min_max_normalized_data(self, query_instance):
            a dataframe, a list, or a list of dicts"""
         query_instance = self.prepare_query_instance(query_instance)
         ohe_base_df = self.prepare_df_for_ohe_encoding()
-        temp = ohe_base_df.append(query_instance, ignore_index=True, sort=False)
+        temp = pd.concat([ohe_base_df, query_instance], ignore_index=True, sort=False)
         temp = self.one_hot_encode_data(temp)
         temp = temp.tail(query_instance.shape[0]).reset_index(drop=True)
         # returns a pandas dataframe

dice_ml/data_interfaces/public_data_interface.py

@@ -6,11 +6,11 @@
 
 import numpy as np
 import pandas as pd
+from raiutils.exceptions import UserConfigValidationException
 from sklearn.preprocessing import LabelEncoder
 
 from dice_ml.data_interfaces.base_data_interface import _BaseData
-from dice_ml.utils.exception import (SystemException,
-                                     UserConfigValidationException)
+from dice_ml.utils.exception import SystemException
 
 
 class PublicData(_BaseData):
@@ -138,9 +138,11 @@ def get_features_range(self, permitted_range_input=None, features_dict=None):
 
     def get_data_type(self, col):
         """Infers data type of a continuous feature from the training data."""
-        if (self.data_df[col].dtype == np.int64) or (self.data_df[col].dtype == np.int32):
+        if (self.data_df[col].dtype == np.int64) or (self.data_df[col].dtype == np.int32) or \
+                (self.data_df[col].dtype == np.int16) or (self.data_df[col].dtype == np.int8):
             return 'int'
-        elif (self.data_df[col].dtype == np.float64) or (self.data_df[col].dtype == np.float32):
+        elif (self.data_df[col].dtype == np.float64) or (self.data_df[col].dtype == np.float32) or \
+                (self.data_df[col].dtype == np.float16):
             return 'float'
         else:
             raise ValueError("Unknown data type of feature %s: must be int or float" % col)

dice_ml/dice.py

@@ -2,10 +2,11 @@
    frameworks such as Tensorflow or PyTorch or sklearn, and different methods
    such as RandomSampling, DiCEKD or DiCEGenetic"""
 
+from raiutils.exceptions import UserConfigValidationException
+
 from dice_ml.constants import BackEndTypes, SamplingStrategy
 from dice_ml.data_interfaces.private_data_interface import PrivateData
 from dice_ml.explainer_interfaces.explainer_base import ExplainerBase
-from dice_ml.utils.exception import UserConfigValidationException
 
 
 class Dice(ExplainerBase):

dice_ml/diverse_counterfactuals.py

@@ -1,5 +1,6 @@
 import copy
 import json
+import math
 
 import pandas as pd
 
@@ -134,10 +135,16 @@ def display_df(self, df, show_only_changes):
             org = self.test_instance_df.values.tolist()[0]
             for ix in range(df.shape[0]):
                 for jx in range(len(org)):
-                    if newdf[ix][jx] == org[jx]:
-                        newdf[ix][jx] = '-'
+                    if not isinstance(newdf[ix][jx], str):
+                        if math.isclose(newdf[ix][jx], org[jx], rel_tol=abs(org[jx]/10000)):
+                            newdf[ix][jx] = '-'
+                        else:
+                            newdf[ix][jx] = str(newdf[ix][jx])
                     else:
-                        newdf[ix][jx] = str(newdf[ix][jx])
+                        if newdf[ix][jx] == org[jx]:
+                            newdf[ix][jx] = '-'
+                        else:
+                            newdf[ix][jx] = str(newdf[ix][jx])
             display(pd.DataFrame(newdf, columns=df.columns, index=df.index))  # works only in Jupyter notebook
 
     def visualize_as_list(self, display_sparse_df=True, show_only_changes=False):
@@ -157,7 +164,7 @@ def print_list(self, li, show_only_changes):
             org = self.test_instance_df.values.tolist()[0]
             for ix in range(len(newli)):
                 for jx in range(len(newli[ix])):
-                    if newli[ix][jx] == org[jx]:
+                    if math.isclose(newli[ix][jx], org[jx], rel_tol=abs(org[jx]/10000)):
                         newli[ix][jx] = '-'
                 print(newli[ix])
 

dice_ml/explainer_interfaces/dice_genetic.py

@@ -8,11 +8,11 @@
 
 import numpy as np
 import pandas as pd
+from raiutils.exceptions import UserConfigValidationException
 
 from dice_ml import diverse_counterfactuals as exp
 from dice_ml.constants import ModelTypes
 from dice_ml.explainer_interfaces.explainer_base import ExplainerBase
-from dice_ml.utils.exception import UserConfigValidationException
 
 
 class DiceGenetic(ExplainerBase):
@@ -507,7 +507,7 @@ def find_counterfactuals(self, query_instance, desired_range, desired_class,
 
         # converting to dataframe
         query_instance_df = self.label_decode(query_instance)
-        query_instance_df[self.data_interface.outcome_name] = self.test_pred
+        query_instance_df[self.data_interface.outcome_name] = self.get_model_output_from_scores(self.test_pred)
         self.final_cfs_df = self.label_decode_cfs(self.final_cfs)
         self.final_cfs_df_sparse = copy.deepcopy(self.final_cfs_df)
 

dice_ml/explainer_interfaces/dice_pytorch.py

@@ -508,7 +508,7 @@ def find_counterfactuals(self, query_instance, desired_class, optimizer, learnin
                 temp_cfs_stored = self.round_off_cfs(assign=False)
                 test_preds_stored = [self.predict_fn(cf) for cf in temp_cfs_stored]
 
-                if((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in test_preds_stored)) or
+                if ((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in test_preds_stored)) or
                    (self.target_cf_class == 1 and all(i >= self.stopping_threshold for i in test_preds_stored))):
                     avg_preds_dist = np.mean([abs(pred[0]-self.stopping_threshold) for pred in test_preds_stored])
                     if avg_preds_dist < self.min_dist_from_threshold[loop_ix]:
@@ -533,7 +533,7 @@ def find_counterfactuals(self, query_instance, desired_class, optimizer, learnin
         self.cfs_preds = [self.predict_fn(cfs) for cfs in self.final_cfs]
 
         # update final_cfs from backed up CFs if valid CFs are not found
-        if((self.target_cf_class == 0 and any(i[0] > self.stopping_threshold for i in self.cfs_preds)) or
+        if ((self.target_cf_class == 0 and any(i[0] > self.stopping_threshold for i in self.cfs_preds)) or
            (self.target_cf_class == 1 and any(i[0] < self.stopping_threshold for i in self.cfs_preds))):
             for loop_ix in range(loop_find_CFs):
                 if self.min_dist_from_threshold[loop_ix] != 100:
@@ -580,7 +580,7 @@ def find_counterfactuals(self, query_instance, desired_class, optimizer, learnin
             final_cfs_df_sparse = None
 
         m, s = divmod(self.elapsed, 60)
-        if((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in self.cfs_preds)) or
+        if ((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in self.cfs_preds)) or
            (self.target_cf_class == 1 and all(i >= self.stopping_threshold for i in self.cfs_preds))):
             self.total_CFs_found = max(loop_find_CFs, self.total_CFs)
             valid_ix = [ix for ix in range(max(loop_find_CFs, self.total_CFs))]  # indexes of valid CFs
@@ -590,7 +590,7 @@ def find_counterfactuals(self, query_instance, desired_class, optimizer, learnin
             self.total_CFs_found = 0
             valid_ix = []  # indexes of valid CFs
             for cf_ix, pred in enumerate(self.cfs_preds):
-                if((self.target_cf_class == 0 and pred[0][0] < self.stopping_threshold) or
+                if ((self.target_cf_class == 0 and pred[0][0] < self.stopping_threshold) or
                    (self.target_cf_class == 1 and pred[0][0] > self.stopping_threshold)):
                     self.total_CFs_found += 1
                     valid_ix.append(cf_ix)

dice_ml/explainer_interfaces/dice_tensorflow1.py

@@ -609,7 +609,7 @@ def find_counterfactuals(self, query_instance, limit_steps_ls, desired_class="op
                 temp_cfs_stored = self.round_off_cfs(assign=False)
                 test_preds_stored = [self.predict_fn(cf) for cf in temp_cfs_stored]
 
-                if((self.target_cf_class[0][0] == 0 and all(i <= self.stopping_threshold for i in test_preds_stored)) or
+                if ((self.target_cf_class[0][0] == 0 and all(i <= self.stopping_threshold for i in test_preds_stored)) or
                    (self.target_cf_class[0][0] == 1 and all(i >= self.stopping_threshold for i in test_preds_stored))):
                     avg_preds_dist = np.mean([abs(pred-self.stopping_threshold) for pred in test_preds_stored])
                     if avg_preds_dist < self.min_dist_from_threshold[loop_ix]:
@@ -634,7 +634,7 @@ def find_counterfactuals(self, query_instance, limit_steps_ls, desired_class="op
         self.cfs_preds = [self.predict_fn(cfs) for cfs in self.final_cfs]
 
         # update final_cfs from backed up CFs if valid CFs are not found
-        if((self.target_cf_class == 0 and any(i[0] > self.stopping_threshold for i in self.cfs_preds)) or
+        if ((self.target_cf_class == 0 and any(i[0] > self.stopping_threshold for i in self.cfs_preds)) or
            (self.target_cf_class == 1 and any(i[0] < self.stopping_threshold for i in self.cfs_preds))):
             for loop_ix in range(loop_find_CFs):
                 if self.min_dist_from_threshold[loop_ix] != 100:
@@ -666,7 +666,7 @@ def find_counterfactuals(self, query_instance, limit_steps_ls, desired_class="op
             final_cfs_df_sparse = None
 
         m, s = divmod(self.elapsed, 60)
-        if((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in self.cfs_preds)) or
+        if ((self.target_cf_class == 0 and all(i <= self.stopping_threshold for i in self.cfs_preds)) or
            (self.target_cf_class == 1 and all(i >= self.stopping_threshold for i in self.cfs_preds))):
             self.total_CFs_found = max(loop_find_CFs, self.total_CFs)
             valid_ix = [ix for ix in range(max(loop_find_CFs, self.total_CFs))]  # indexes of valid CFs
@@ -676,7 +676,7 @@ def find_counterfactuals(self, query_instance, limit_steps_ls, desired_class="op
             self.total_CFs_found = 0
             valid_ix = []  # indexes of valid CFs
             for cf_ix, pred in enumerate(self.cfs_preds):
-                if((self.target_cf_class == 0 and pred < self.stopping_threshold) or
+                if ((self.target_cf_class == 0 and pred < self.stopping_threshold) or
                    (self.target_cf_class == 1 and pred > self.stopping_threshold)):
                     self.total_CFs_found += 1
                     valid_ix.append(cf_ix)