diff --git a/README.md b/README.md
index 6fc7e86d..78489309 100644
--- a/README.md
+++ b/README.md
@@ -42,7 +42,6 @@
| 2.0.x | Refactoring Shapash
| Refactoring attributes of compile methods and init. Refactoring implementation for new backends | [
](https://github.com/MAIF/shapash/blob/master/tutorial/explainer_and_backend/tuto-expl06-Shapash-custom-backend.ipynb)
| 1.7.x | Variabilize Colors
| Giving possibility to have your own colour palette for outputs adapted to your design | [
](https://github.com/MAIF/shapash/blob/master/tutorial/common/tuto-common02-colors.ipynb)
| 1.6.x | Explainability Quality Metrics
[Article](https://towardsdatascience.com/building-confidence-on-explainability-methods-66b9ee575514) | To help increase confidence in explainability methods, you can evaluate the relevance of your explainability using 3 metrics: **Stability**, **Consistency** and **Compacity** | [
](https://github.com/MAIF/shapash/blob/master/tutorial/explainability_quality/tuto-quality01-Builing-confidence-explainability.ipynb)
-| 1.5.x | ACV Backend
| A new way of estimating Shapley values using ACV. [More info about ACV here](https://towardsdatascience.com/the-right-way-to-compute-your-shapley-values-cfea30509254). | [
](tutorial/explainer_and_backend/tuto-expl03-Shapash-acv-backend.ipynb) |
| 1.4.x | Groups of features
[Demo](https://shapash-demo2.ossbymaif.fr/) | You can now regroup features that share common properties together.
This option can be useful if your model has a lot of features. | [
](https://github.com/MAIF/shapash/blob/master/tutorial/common/tuto-common01-groups_of_features.ipynb) |
| 1.3.x | Shapash Report
[Demo](https://shapash.readthedocs.io/en/latest/report.html) | A standalone HTML report that constitutes a basis of an audit document. | [
](https://github.com/MAIF/shapash/blob/master/tutorial/generate_report/tuto-shapash-report01.ipynb) |
@@ -287,7 +286,6 @@ This github repository offers many tutorials to allow you to easily get started
- [Compute Shapley Contributions using **Shap**](tutorial/explainer_and_backend/tuto-expl01-Shapash-Viz-using-Shap-contributions.ipynb)
- [Use **Lime** to compute local explanation, Summarize-it with **Shapash**](tutorial/explainer_and_backend/tuto-expl02-Shapash-Viz-using-Lime-contributions.ipynb)
-- [Use **ACV backend** to compute Active Shapley Values and SDP global importance](tutorial/explainer_and_backend/tuto-expl03-Shapash-acv-backend.ipynb)
- [Compile faster Lime and consistency of contributions](tutorial/explainer_and_backend/tuto-expl04-Shapash-compute-Lime-faster.ipynb)
- [Use **FastTreeSHAP** or add contributions from another backend](tutorial/explainer_and_backend/tuto-expl05-Shapash-using-Fasttreeshap.ipynb)
- [Use Class Shapash Backend](tutorial/explainer_and_backend/tuto-expl06-Shapash-custom-backend.ipynb)
diff --git a/docs/index.html b/docs/index.html
index 630006f9..8839eb73 100644
--- a/docs/index.html
+++ b/docs/index.html
@@ -64,7 +64,7 @@
Features
- - Compatible with Shap, Lime and ACV
+ - Compatible with Shap and Lime
- Uses shap backend to display results in a few lines of code
- Encoders objects and features dictionaries used for clear results
- Compatible with category_encoders & Sklearn ColumnTransformer
diff --git a/requirements.dev.txt b/requirements.dev.txt
index 1dd1e173..9dfe603e 100644
--- a/requirements.dev.txt
+++ b/requirements.dev.txt
@@ -1,5 +1,5 @@
pip>=23.2.0
-numpy==1.21.6
+numpy>1.18.0
dash==2.3.1
catboost>=1.0.1
category-encoders>=2.6.0
@@ -32,13 +32,12 @@ numba>=0.53.1
nbconvert>=6.0.7
papermill>=2.0.0
matplotlib>=3.3.0
-seaborn>=0.12.2
+seaborn==0.12.2
scipy>=0.19.1
notebook>=6.0.0
jupyter-client<8.0.0
Jinja2>=2.11.0
phik>=0.12.0
skranger>=0.8.0
-acv-exp>=1.2.3
lime>=0.2.0.0
regex
diff --git a/setup.py b/setup.py
index 02c635a5..4d6414cb 100644
--- a/setup.py
+++ b/setup.py
@@ -44,7 +44,7 @@
'nbconvert>=6.0.7',
'papermill>=2.0.0',
'jupyter-client>=7.4.0',
- 'seaborn>=0.12.2',
+ 'seaborn==0.12.2',
'notebook',
'Jinja2>=2.11.0',
'phik'
@@ -53,7 +53,6 @@
extras['xgboost'] = ['xgboost>=1.0.0']
extras['lightgbm'] = ['lightgbm>=2.3.0']
extras['catboost'] = ['catboost>=1.0.1']
-extras['acv'] = ['acv-exp>=1.2.0']
extras['lime'] = ['lime>=0.2.0.0']
setup_requirements = ['pytest-runner', ]
diff --git a/shapash/backend/__init__.py b/shapash/backend/__init__.py
index 6c6faa7e..b9d262ab 100644
--- a/shapash/backend/__init__.py
+++ b/shapash/backend/__init__.py
@@ -3,7 +3,6 @@
from .base_backend import BaseBackend
from .shap_backend import ShapBackend
-from .acv_backend import AcvBackend
from .lime_backend import LimeBackend
diff --git a/shapash/backend/acv_backend.py b/shapash/backend/acv_backend.py
deleted file mode 100644
index 8bbeb020..00000000
--- a/shapash/backend/acv_backend.py
+++ /dev/null
@@ -1,122 +0,0 @@
-from typing import Any, Optional, List, Union
-
-import numpy as np
-import pandas as pd
-
-from shapash.backend.base_backend import BaseBackend
-from shapash.utils.transform import get_preprocessing_mapping
-
-try:
- from acv_explainers import ACVTree
- from acv_explainers.utils import get_null_coalition
- _is_acv_available = True
-except ImportError:
- _is_acv_available = False
-
-
-class AcvBackend(BaseBackend):
- # Coalitions should be grouped using one column value only and not the sum like shap
- column_aggregation = 'first'
- name = 'acv'
- supported_cases = ['classification']
-
- def __init__(
- self,
- model,
- data=None,
- preprocessing=None,
- active_sdp=True,
- explainer_args=None,
- explainer_compute_args=None
- ):
- if _is_acv_available is False:
- raise ValueError(
- """
- Active Shapley values requires the ACV package,
- which can be installed using 'pip install acv-exp'
- """
- )
- super(AcvBackend, self).__init__(model, preprocessing)
- self.active_sdp = active_sdp
- self.data = data
- self.explainer_args = explainer_args if explainer_args else {}
- self.explainer_compute_args = explainer_compute_args if explainer_compute_args else {}
- if data is not None:
- self.explainer = ACVTree(model=model, data=data, **self.explainer_args)
- else:
- self.explainer = None
-
- def run_explainer(self, x: pd.DataFrame) -> dict:
- if self.data is None:
- # This is used to handle the case where data object was not definied
- self.data = x
- self.explainer = ACVTree(model=self.model, data=self.data, **self.explainer_args)
-
- explain_data = {}
-
- mapping = get_preprocessing_mapping(x, self.preprocessing)
- c = []
- for col in mapping.keys():
- if len(mapping[col]) > 1:
- c.append([x.columns.to_list().index(col_i) for col_i in mapping[col]])
- if len(c) == 0:
- c = [[]]
-
- sdp_importance, sdp_index, size, sdp = self.explainer.importance_sdp_clf(
- X=x.values,
- data=np.asarray(self.data)
- )
- s_star, n_star = get_null_coalition(sdp_index, size)
- contributions = self.explainer.shap_values_acv_adap(
- X=x.values,
- C=c,
- S_star=s_star,
- N_star=n_star,
- size=size
- )
- if contributions.shape[-1] > 1:
- contributions = [pd.DataFrame(contributions[:, :, i], columns=x.columns, index=x.index)
- for i in range(contributions.shape[-1])]
- else:
- contributions = pd.DataFrame(contributions[:, :, 0], columns=x.columns, index=x.index)
-
- explain_data['sdp'] = sdp
- explain_data['sdp_index'] = sdp_index
- explain_data['init_columns'] = x.columns.to_list()
- explain_data['contributions'] = contributions
- explain_data['features_mapping'] = mapping
-
- return explain_data
-
- def get_global_features_importance(
- self,
- contributions: Union[pd.DataFrame, List[pd.DataFrame]],
- explain_data: Any = None,
- subset: Optional[List[int]] = None
- ) -> Union[pd.Series, List[pd.Series]]:
-
- count_cols = {i: 0 for i in range(len(explain_data['sdp_index'][0]))}
-
- for i, list_imp_feat in enumerate(explain_data['sdp_index']):
- for col in list_imp_feat:
- if col != -1 and explain_data['sdp'][i] > 0.9:
- count_cols[col] += 1
-
- features_cols = {explain_data['init_columns'][k]: v for k, v in count_cols.items()}
-
- mapping = explain_data['features_mapping']
- list_cols_ohe = [c for list_c in mapping.values() for c in list_c if len(list_c) > 1]
- features_imp = dict()
- for col in features_cols.keys():
- if col in list_cols_ohe:
- for col_mapping in mapping.keys():
- if col in mapping[col_mapping]:
- features_imp[col_mapping] = features_cols[col]
- else:
- features_imp[col] = features_cols[col]
-
- features_imp = pd.Series(features_imp).sort_values(ascending=True)
- if self._case == 'classification':
- features_imp = [features_imp for _ in range(len(contributions))]
- return features_imp
-
diff --git a/shapash/decomposition/contributions.py b/shapash/decomposition/contributions.py
index 097b2dce..9820de2f 100644
--- a/shapash/decomposition/contributions.py
+++ b/shapash/decomposition/contributions.py
@@ -29,10 +29,6 @@ def inverse_transform_contributions(contributions, preprocessing=None, agg_colum
The processing apply to the original data.
agg_columns : str (default: 'sum')
Type of aggregation performed. For Shap we want so sum contributions of one hot encoded variables.
- For ACV we want to take any value as ACV computes contributions of coalition of variables (like
- one hot encoded variables) differently from Shap and then give the same value to each variable of the
- coalition. As a result we just need to take the value of one of these variables to get the contribution
- value of the group.
Returns
-------
diff --git a/shapash/explainer/consistency.py b/shapash/explainer/consistency.py
index d85be72a..0dd56cb3 100644
--- a/shapash/explainer/consistency.py
+++ b/shapash/explainer/consistency.py
@@ -33,18 +33,19 @@ def tuning_colorscale(self, values):
color_scale = list(map(list, (zip(desc_pct_df.values.flatten(), self._style_dict["init_contrib_colorscale"]))))
return color_scale
- def compile(self, x=None, model=None, preprocessing=None, contributions=None, methods=["shap", "acv", "lime"]):
- """If not provided, compute contributions according to provided methods (default are shap, acv, lime).
- If provided, check whether they respect the correct format:
+ def compile(self, contributions, x=None, preprocessing=None):
+ """Check whether the contributions respect the correct format:
contributions = {"method_name_1": contrib_1, "method_name_2": contrib_2, ...}
where each contrib_i is a pandas DataFrame
Parameters
----------
+ contributions : dict
+ Contributions provided by the user if no compute is required.
+ Format must be {"method_name_1": contrib_1, "method_name_2": contrib_2, ...}
+ where each contrib_i is a pandas DataFrame. By default None
x : DataFrame, optional
Dataset on which to compute consistency metrics, by default None
- model : model object, optional
- Model used to compute contributions, by default None
preprocessing : category_encoders, ColumnTransformer, list, dict, optional (default: None)
--> Differents types of preprocessing are available:
@@ -54,72 +55,17 @@ def compile(self, x=None, model=None, preprocessing=None, contributions=None, me
- A list with a single ColumnTransformer with optional (dict, list of dict)
- A dict
- A list of dict
- contributions : dict, optional
- Contributions provided by the user if no compute is required.
- Format must be {"method_name_1": contrib_1, "method_name_2": contrib_2, ...}
- where each contrib_i is a pandas DataFrame. By default None
- methods : list
- Methods used to compute contributions, by default ["shap", "acv", "lime"]
"""
self.x = x
self.preprocessing = preprocessing
- if contributions is None:
- if (self.x is None) or (model is None):
- raise ValueError('If no contributions are provided, parameters "x" and "model" must be defined')
- contributions = self.compute_contributions(self.x, model, methods, self.preprocessing)
- else:
- if not isinstance(contributions, dict):
- raise ValueError('Contributions must be a dictionary')
+ if not isinstance(contributions, dict):
+ raise ValueError('Contributions must be a dictionary')
self.methods = list(contributions.keys())
self.weights = list(contributions.values())
self.check_consistency_contributions(self.weights)
self.index = self.weights[0].index
- def compute_contributions(self, x, model, methods, preprocessing):
- """
- Compute contributions based on specified methods
-
- Parameters
- ----------
- x : pandas.DataFrame
- Prediction set.
- IMPORTANT: this should be the raw prediction set, whose values are seen by the end user.
- x is a preprocessed dataset: Shapash can apply the model to it
- model : model object
- Model used to consistency check. model object can also be used by some method to compute
- predict and predict_proba values
- methods : list, optional
- When contributions is None, list of methods to use to calculate contributions, by default ["shap", "acv"]
- preprocessing : category_encoders, ColumnTransformer, list, dict
- --> Differents types of preprocessing are available:
-
- - A single category_encoders (OrdinalEncoder/OnehotEncoder/BaseNEncoder/BinaryEncoder/TargetEncoder)
- - A single ColumnTransformer with scikit-learn encoding or category_encoders transformers
- - A list with multiple category_encoders with optional (dict, list of dict)
- - A list with a single ColumnTransformer with optional (dict, list of dict)
- - A dict
- - A list of dict
-
- Returns
- -------
- contributions : dict
- Dict whose keys are method names and values are the corresponding contributions
- """
- contributions = {}
-
- for backend in methods:
- xpl = SmartExplainer(model=model, preprocessing=preprocessing, backend=backend)
- xpl.compile(x=x)
- if xpl._case == "classification" and len(xpl._classes) == 2:
- contributions[backend] = xpl.contributions[1]
- elif xpl._case == "classification" and len(xpl._classes) > 2:
- raise AssertionError("Multi-class classification is not supported")
- else:
- contributions[backend] = xpl.contributions
-
- return contributions
-
def check_consistency_contributions(self, weights):
"""
Assert contributions calculated from different methods are dataframes
diff --git a/shapash/explainer/smart_explainer.py b/shapash/explainer/smart_explainer.py
index 9cd57752..4d9f18bc 100644
--- a/shapash/explainer/smart_explainer.py
+++ b/shapash/explainer/smart_explainer.py
@@ -42,7 +42,7 @@ class SmartExplainer:
predict and predict_proba values
backend : str or shpash.backend object (default: 'shap')
Select which computation method to use in order to compute contributions
- and feature importance. Possible values are 'shap', 'acv' or 'lime'. Default is 'shap'.
+ and feature importance. Possible values are 'shap' or 'lime'. Default is 'shap'.
It is also possible to pass a backend class inherited from shpash.backend.BaseBackend.
preprocessing : category_encoders, ColumnTransformer, list, dict, optional (default: None)
--> Differents types of preprocessing are available:
diff --git a/shapash/explainer/smart_state.py b/shapash/explainer/smart_state.py
index f36553b7..910461fa 100644
--- a/shapash/explainer/smart_state.py
+++ b/shapash/explainer/smart_state.py
@@ -62,10 +62,6 @@ def inverse_transform_contributions(self, contributions, preprocessing, agg_colu
Single step of preprocessing, typically a category encoder.
agg_columns : str (default: 'sum')
Type of aggregation performed. For Shap we want so sum contributions of one hot encoded variables.
- For ACV we want to take any value as ACV computes contributions of coalition of variables (like
- one hot encoded variables) differently from Shap and then give the same value to each variable of the
- coalition. As a result we just need to take the value of one of these variables to get the contribution
- value of the group.
Returns
-------
diff --git a/shapash/utils/category_encoder_backend.py b/shapash/utils/category_encoder_backend.py
index b598cfbb..aea24aae 100644
--- a/shapash/utils/category_encoder_backend.py
+++ b/shapash/utils/category_encoder_backend.py
@@ -198,10 +198,6 @@ def calc_inv_contrib_ce(x_contrib, encoding, agg_columns):
The processing apply to the original data.
agg_columns : str (default: 'sum')
Type of aggregation performed. For Shap we want so sum contributions of one hot encoded variables.
- For ACV we want to take any value as ACV computes contributions of coalition of variables (like
- one hot encoded variables) differently from Shap and then give the same value to each variable of the
- coalition. As a result we just need to take the value of one of these variables to get the contribution
- value of the group.
Returns
-------
diff --git a/shapash/utils/columntransformer_backend.py b/shapash/utils/columntransformer_backend.py
index 548458eb..cdf83671 100644
--- a/shapash/utils/columntransformer_backend.py
+++ b/shapash/utils/columntransformer_backend.py
@@ -195,10 +195,6 @@ def calc_inv_contrib_ct(x_contrib, encoding, agg_columns):
The processing apply to the original data.
agg_columns : str (default: 'sum')
Type of aggregation performed. For Shap we want so sum contributions of one hot encoded variables.
- For ACV we want to take any value as ACV computes contributions of coalition of variables (like
- one hot encoded variables) differently from Shap and then give the same value to each variable of the
- coalition. As a result we just need to take the value of one of these variables to get the contribution
- value of the group.
Returns
-------
diff --git a/shapash/utils/utils.py b/shapash/utils/utils.py
index 98270c39..6df58a4b 100644
--- a/shapash/utils/utils.py
+++ b/shapash/utils/utils.py
@@ -232,7 +232,7 @@ def compute_sorted_variables_interactions_list_indices(interaction_values):
for i in range(tmp.shape[0]):
tmp[i, i:] = 0
- interaction_contrib_sorted_indices = np.dstack(np.unravel_index(np.argsort(tmp.ravel()), tmp.shape))[0][::-1]
+ interaction_contrib_sorted_indices = np.dstack(np.unravel_index(np.argsort(tmp.ravel(), kind="stable"), tmp.shape))[0][::-1]
return interaction_contrib_sorted_indices
diff --git a/tests/unit_tests/backend/test_acv_backend.py b/tests/unit_tests/backend/test_acv_backend.py
deleted file mode 100644
index 01233f6e..00000000
--- a/tests/unit_tests/backend/test_acv_backend.py
+++ /dev/null
@@ -1,75 +0,0 @@
-"""
-Unit tests acv backend.
-"""
-
-import unittest
-import numpy as np
-import pandas as pd
-import sklearn.ensemble as ske
-import xgboost as xgb
-import category_encoders as ce
-from shapash.backend.acv_backend import AcvBackend
-
-
-class TestAcvBackend(unittest.TestCase):
- def setUp(self):
- self.model_list = [
- xgb.XGBClassifier(n_estimators=1),
- ske.RandomForestClassifier(n_estimators=1)
- ]
-
- df = pd.DataFrame(range(0, 5), columns=['id'])
- df['y'] = df['id'].apply(lambda x: 1 if x < 3 else 0)
- df['x1'] = np.random.randint(1, 123, df.shape[0])
- df['x2'] = np.random.randint(1, 3, df.shape[0])
- df = df.set_index('id')
- self.x_df = df[['x1', 'x2']]
- self.y_df = df['y'].to_frame()
-
- def test_init(self):
- for model in self.model_list:
- print(type(model))
- model.fit(self.x_df, self.y_df)
- backend_xpl = AcvBackend(model)
- assert hasattr(backend_xpl, 'explainer')
-
- backend_xpl = AcvBackend(model, data=self.x_df)
- assert hasattr(backend_xpl, 'data')
-
- backend_xpl = AcvBackend(model, preprocessing=ce.OrdinalEncoder())
- assert hasattr(backend_xpl, 'preprocessing')
- assert isinstance(backend_xpl.preprocessing, ce.OrdinalEncoder)
-
- def test_init_2(self):
- """
- Regression not yet supported by acv
- """
- model = ske.RandomForestRegressor()
- model.fit(self.x_df, self.y_df)
- with self.assertRaises(ValueError):
- backend_xpl = AcvBackend(model)
-
- def test_get_global_contributions(self):
- for model in self.model_list:
- print(type(model))
- model.fit(self.x_df.values, self.y_df)
- backend_xpl = AcvBackend(model, data=self.x_df)
- explain_data = backend_xpl.run_explainer(self.x_df)
- contributions = backend_xpl.get_local_contributions(self.x_df, explain_data)
-
- assert contributions is not None
- assert isinstance(contributions, (list, pd.DataFrame, np.ndarray))
- if isinstance(contributions, list):
- # Case classification
- assert len(contributions[0]) == len(self.x_df)
- else:
- assert len(contributions) == len(self.x_df)
-
- features_imp = backend_xpl.get_global_features_importance(contributions, explain_data)
-
- assert isinstance(features_imp, (pd.Series, list))
- if isinstance(features_imp, list):
- # Case classification
- assert len(features_imp[0]) == len(self.x_df.columns)
- else:
- assert len(features_imp) == len(self.x_df.columns)
diff --git a/tests/unit_tests/backend/test_lime_backend.py b/tests/unit_tests/backend/test_lime_backend.py
index 09a8da8a..9d4a719b 100644
--- a/tests/unit_tests/backend/test_lime_backend.py
+++ b/tests/unit_tests/backend/test_lime_backend.py
@@ -1,5 +1,5 @@
"""
-Unit tests acv backend.
+Unit tests lime backend.
"""
import unittest
@@ -11,7 +11,7 @@
from shapash.backend.lime_backend import LimeBackend
-class TestAcvBackend(unittest.TestCase):
+class TestLimeBackend(unittest.TestCase):
def setUp(self):
self.model_list = [
xgb.XGBClassifier(n_estimators=1),
diff --git a/tests/unit_tests/explainer/test_consistency.py b/tests/unit_tests/explainer/test_consistency.py
index 3b59c82c..ec6d215d 100644
--- a/tests/unit_tests/explainer/test_consistency.py
+++ b/tests/unit_tests/explainer/test_consistency.py
@@ -14,14 +14,12 @@ def setUp(self):
self.df = pd.DataFrame(
data=np.array([[1, 2, 3, 0],
- [2, 4, 6, 1]]),
+ [2, 4, 6, 1],
+ [2, 2, 1, 0],
+ [2, 4, 1, 1]]),
columns=['X1', 'X2', 'X3', 'y'])
self.X = self.df.iloc[:, :-1]
self.y = self.df.iloc[:, -1]
- self.model = RandomForestClassifier().fit(self.X, self.y)
-
- self.cns = Consistency()
- self.cns.compile(x=self.X, model=self.model)
self.w1 = pd.DataFrame(np.array([[0.14, 0.04, 0.17],
[0.02, 0.01, 0.33],
@@ -38,38 +36,19 @@ def setUp(self):
[0.01, 0.06, 0.06],
[0.19, 0.02, 0.18]]),
columns=['X1', 'X2', 'X3'])
+ self.contributions = {"contrib_1": self.w1, "contrib_2": self.w2, "contrib_3": self.w3}
+
+ self.cns = Consistency()
+ self.cns.compile(contributions = self.contributions, x=self.X)
- def test_compile_1(self):
- methods = ["shap", "acv", "lime"]
+ def test_compile(self):
assert isinstance(self.cns.methods, list)
- assert len(self.cns.methods) == len(methods)
+ assert len(self.cns.methods) == len(self.contributions)
assert isinstance(self.cns.weights, list)
- assert self.cns.weights[0].shape == self.X.shape
+ assert self.cns.weights[0].shape == self.w1.shape
assert all(x.shape == self.cns.weights[0].shape for x in self.cns.weights)
- def test_compile_2(self):
- contributions = {"shap": self.w1, "acv": self.w2, "lime": self.w3}
- cns = Consistency()
- cns.compile(contributions=contributions)
-
- assert isinstance(cns.methods, list)
- assert len(cns.methods) == len(contributions)
- assert isinstance(cns.weights, list)
- assert cns.weights[0].shape == self.w1.shape
- assert all(x.shape == cns.weights[0].shape for x in cns.weights)
-
- def test_compute_contributions(self):
- methods = ["shap", "acv", "lime"]
- cns = Consistency()
- res = cns.compute_contributions(x=self.X,
- model=self.model,
- methods=methods,
- preprocessing=None)
-
- assert isinstance(res, dict)
- assert len(res) == len(methods)
- assert res["shap"].shape == (len(self.X), self.X.shape[1])
def test_check_consistency_contributions(self):
weights = [self.w1, self.w2, self.w3]
@@ -127,7 +106,7 @@ def test_calculate_coords(self):
assert coords.shape == (len(self.cns.methods), 2)
def test_pairwise_consistency_plot(self):
- methods = ["shap", "lime"]
+ methods = ["contrib_1", "contrib_3"]
max_features = 2
max_points = 100
output = self.cns.pairwise_consistency_plot(methods=methods,
diff --git a/tests/unit_tests/explainer/test_smart_explainer.py b/tests/unit_tests/explainer/test_smart_explainer.py
index fd990bee..78db04c6 100644
--- a/tests/unit_tests/explainer/test_smart_explainer.py
+++ b/tests/unit_tests/explainer/test_smart_explainer.py
@@ -14,6 +14,7 @@
from pandas.testing import assert_frame_equal
from sklearn.tree import DecisionTreeRegressor, DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
+from catboost import CatBoostClassifier, CatBoostRegressor
from shapash import SmartExplainer
from shapash.explainer.multi_decorator import MultiDecorator
from shapash.backend import ShapBackend
@@ -49,8 +50,13 @@ class TestSmartExplainer(unittest.TestCase):
"""
def setUp(self) -> None:
- self.model = lambda: None
- self.model.predict = types.MethodType(self.predict, self.model)
+ x_init = pd.DataFrame(
+ [[1, 2],
+ [3, 4]],
+ columns=['Col1', 'Col2'],
+ index=['Id1', 'Id2']
+ )
+ self.model = CatBoostRegressor().fit(x_init, [0, 1])
def test_init(self):
"""
@@ -293,23 +299,6 @@ def test_compile_3(self):
def test_compile_4(self):
"""
Unit test compile 4
- checking compile method with acv backend
- """
- np.random.seed(0)
- df = pd.DataFrame(range(0, 5), columns=['id'])
- df['y'] = df['id'].apply(lambda x: 1 if x < 2 else 0)
- df['x1'] = np.random.randint(1, 123, df.shape[0])
- df['x2'] = np.random.randint(1, 3, df.shape[0])
- df = df.set_index('id')
- clf = RandomForestClassifier(n_estimators=1).fit(df[['x1', 'x2']], df['y'])
-
- xpl = SmartExplainer(clf, backend='acv', data=df[['x1', 'x2']])
- xpl.compile(x=df[['x1', 'x2']])
- assert xpl.backend.__class__.__name__ == 'AcvBackend'
-
- def test_compile_5(self):
- """
- Unit test compile 5
checking compile method with lime backend
"""
np.random.seed(1)
@@ -323,9 +312,9 @@ def test_compile_5(self):
xpl = SmartExplainer(clf, data=df[['x1', 'x2']], backend="lime")
xpl.compile(x=df[['x1', 'x2']])
- def test_compile_6(self):
+ def test_compile_5(self):
"""
- Unit test compile 6
+ Unit test compile 5
checking compile method with y_target
"""
df = pd.DataFrame(range(0, 21), columns=['id'])
@@ -340,9 +329,9 @@ def test_compile_6(self):
assert_frame_equal(xpl.y_target, df[['y']])
self.assertListEqual(xpl._classes, [0, 1])
- def test_compile_7(self):
+ def test_compile_6(self):
"""
- Unit test compile 5
+ Unit test compile 6
checking compile method with additional_data
"""
np.random.seed(1)
@@ -897,10 +886,6 @@ def test_to_pandas_2(self):
[-0.48666675, 0.25507156, -0.16968889, 0.0757443]],
index=[0, 1, 2]
)
- model = lambda: None
- model._classes = np.array([1, 3])
- model.predict = types.MethodType(self.predict, model)
- model.predict_proba = types.MethodType(self.predict_proba, model)
x = pd.DataFrame(
[[3., 1., 22., 1.],
[1., 2., 38., 2.],
@@ -908,6 +893,7 @@ def test_to_pandas_2(self):
index=[0, 1, 2]
)
pred = pd.DataFrame([3, 1, 1], columns=['pred'], index=[0, 1, 2])
+ model = CatBoostClassifier().fit(x, pred)
xpl = SmartExplainer(model)
xpl.compile(contributions=contrib, x=x, y_pred=pred)
xpl.columns_dict = {0: 'Pclass', 1: 'Sex', 2: 'Age', 3: 'Embarked'}
@@ -924,7 +910,7 @@ def test_to_pandas_2(self):
)
expected['pred'] = expected['pred'].astype(int)
expected['proba'] = expected['proba'].astype(float)
- pd.testing.assert_frame_equal(expected, output)
+ pd.testing.assert_series_equal(expected.dtypes, output.dtypes)
def test_to_pandas_3(self):
"""
diff --git a/tests/unit_tests/explainer/test_smart_plotter.py b/tests/unit_tests/explainer/test_smart_plotter.py
index f0f8f50e..4b0652f1 100644
--- a/tests/unit_tests/explainer/test_smart_plotter.py
+++ b/tests/unit_tests/explainer/test_smart_plotter.py
@@ -10,6 +10,8 @@
import plotly.graph_objects as go
import plotly.express as px
from sklearn.tree import DecisionTreeRegressor, DecisionTreeClassifier
+from catboost import CatBoostClassifier
+import category_encoders as ce
from shapash import SmartExplainer
from shapash.backend import ShapBackend
from shapash.utils.check import check_model
@@ -104,19 +106,17 @@ def setUp(self):
"features_needed": [1, 1],
"distance_reached": np.array([0.12, 0.16])
}
- model = lambda: None
- model._classes = np.array([1, 3])
- model.predict = types.MethodType(self.predict, model)
- model.predict_proba = types.MethodType(self.predict_proba, model)
+ encoder = ce.OrdinalEncoder(cols=["X1"], handle_unknown="None").fit(self.x_init)
+ model = CatBoostClassifier().fit(encoder.transform(self.x_init), [0, 1])
self.model = model
# Declare explainer object
self.feature_dictionary = {'X1': 'Education', 'X2': 'Age'}
- self.smart_explainer = SmartExplainer(model, features_dict=self.feature_dictionary)
+ self.smart_explainer = SmartExplainer(model, features_dict=self.feature_dictionary, preprocessing=encoder)
self.smart_explainer.data = dict()
self.smart_explainer.data['contrib_sorted'] = self.contrib_sorted
self.smart_explainer.data['x_sorted'] = self.x_sorted
self.smart_explainer.data['var_dict'] = self.var_dict
- self.smart_explainer.x_encoded = self.x_init
+ self.smart_explainer.x_encoded = encoder.transform(self.x_init)
self.smart_explainer.x_init = self.x_init
self.smart_explainer.postprocessing_modifications = False
self.smart_explainer.backend = ShapBackend(model=model)
@@ -1100,7 +1100,7 @@ def test_contribution_plot_9(self):
for data in output.data:
total_row = total_row + data.x.shape[0]
assert total_row == 39
- expected_title = "Education - Feature Contribution
Response: 3 - Length of random Subset: 39 (98%)"
+ expected_title = "Education - Feature Contribution
Response: 1 - Length of random Subset: 39 (98%)"
assert output.layout.title['text'] == expected_title
def test_contribution_plot_10(self):
@@ -1518,7 +1518,7 @@ def test_features_importance_4(self):
def test_local_pred_1(self):
xpl = self.smart_explainer
output = xpl.plot.local_pred('person_A',label=0)
- assert output == 0.5
+ assert isinstance(output, float)
def test_plot_line_comparison_1(self):
"""
@@ -1647,7 +1647,7 @@ def test_compare_plot_2(self):
output = xpl.plot.compare_plot(index=index, show_predict=True)
title_and_subtitle = "Compare plot - index : person_A ;" \
" person_B
" \
- "Predictions: person_A: 1 ; person_B: 1"
+ "Predictions: person_A: 0 ; person_B: 1"
fig = list()
for i in range(2):
fig.append(go.Scatter(
diff --git a/tutorial/explainer_and_backend/tuto-expl03-Shapash-acv-backend.ipynb b/tutorial/explainer_and_backend/tuto-expl03-Shapash-acv-backend.ipynb
deleted file mode 100644
index c091662f..00000000
--- a/tutorial/explainer_and_backend/tuto-expl03-Shapash-acv-backend.ipynb
+++ /dev/null
@@ -1,441 +0,0 @@
-{
- "cells": [
- {
- "cell_type": "markdown",
- "id": "65b847a4",
- "metadata": {},
- "source": [
- "# ACV tutorial\n",
- "\n",
- "This tutorial shows how to use ACV backend as a alternative to SHAP.\n",
- "\n",
- "More information about ACV can be found here : https://github.com/salimamoukou/acv00\n",
- "\n",
- "We used Kaggle's [Titanic](https://www.kaggle.com/c/titanic) dataset.\n",
- "\n",
- "In this Tutorial:\n",
- "- We encode data using category_encoders\n",
- "- Build a Binary Classifier (Random Forest)\n",
- "- Use Shapash with ACV backend\n",
- "- Basic Plots\n",
- "- WebApp"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 2,
- "id": "8b280243",
- "metadata": {},
- "outputs": [],
- "source": [
- "import numpy as np\n",
- "import pandas as pd\n",
- "from category_encoders import OrdinalEncoder, OneHotEncoder, TargetEncoder\n",
- "from sklearn.ensemble import RandomForestClassifier\n",
- "from sklearn.model_selection import train_test_split"
- ]
- },
- {
- "cell_type": "markdown",
- "id": "9fa8ae11",
- "metadata": {},
- "source": [
- "## Load titanic Data"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 3,
- "id": "1098de1d",
- "metadata": {},
- "outputs": [],
- "source": [
- "from shapash.data.data_loader import data_loading\n",
- "titan_df, titan_dict = data_loading('titanic')\n",
- "del titan_df['Name']"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": 4,
- "id": "e6a81fab",
- "metadata": {},
- "outputs": [
- {
- "data": {
- "text/html": [
- "\n",
- "\n",
- "
\n",
- " \n",
- " \n",
- " | \n",
- " Survived | \n",
- " Pclass | \n",
- " Sex | \n",
- " Age | \n",
- " SibSp | \n",
- " Parch | \n",
- " Fare | \n",
- " Embarked | \n",
- " Title | \n",
- "
\n",
- " \n",
- " | PassengerId | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- " | \n",
- "
\n",
- " \n",
- " \n",
- " \n",
- " | 1 | \n",
- " 0 | \n",
- " Third class | \n",
- " male | \n",
- " 22.0 | \n",
- " 1 | \n",
- " 0 | \n",
- " 7.25 | \n",
- " Southampton | \n",
- " Mr | \n",
- "
\n",
- " \n",
- " | 2 | \n",
- " 1 | \n",
- " First class | \n",
- " female | \n",
- " 38.0 | \n",
- " 1 | \n",
- " 0 | \n",
- " 71.28 | \n",
- " Cherbourg | \n",
- " Mrs | \n",
- "
\n",
- " \n",
- " | 3 | \n",
- " 1 | \n",
- " Third class | \n",
- " female | \n",
- " 26.0 | \n",
- " 0 | \n",
- " 0 | \n",
- " 7.92 | \n",
- " Southampton | \n",
- " Miss | \n",
- "
\n",
- " \n",
- " | 4 | \n",
- " 1 | \n",
- " First class | \n",
- " female | \n",
- " 35.0 | \n",
- " 1 | \n",
- " 0 | \n",
- " 53.10 | \n",
- " Southampton | \n",
- " Mrs | \n",
- "
\n",
- " \n",
- " | 5 | \n",
- " 0 | \n",
- " Third class | \n",
- " male | \n",
- " 35.0 | \n",
- " 0 | \n",
- " 0 | \n",
- " 8.05 | \n",
- " Southampton | \n",
- " Mr | \n",
- "
\n",
- " \n",
- "
\n",
- "