remove sklearn shap support to fix dependency issues

Author: tschuelia

pypythia/predictor.py

@@ -112,14 +112,14 @@ def predict(self, query: Dict) -> float:
         return prediction
 
     def plot_shapley_values(self, query: Dict) -> Figure:
+        if isinstance(self.predictor, RandomForestRegressor):
+            raise PyPythiaException("Cannot infer shapley values for scikit-learn predictors")
+
         explainer = shap.TreeExplainer(self.predictor)
         df = self._check_and_pack_query(query)
         shap_values = explainer.shap_values(df)
         base_values = explainer.expected_value
 
-        if isinstance(self.predictor, RandomForestRegressor):
-            base_values = base_values[0]
-
         return shap.plots.waterfall(
             shap.Explanation(
                 values=shap_values[0],

setup.cfg

@@ -13,7 +13,6 @@ classifiers =
     Programming Language :: Python :: 3.9
     Programming Language :: Python :: 3.10
     Programming Language :: Python :: 3.11
-    
 
 [options]
 include_package_data = true

tests/test_predictor.py

@@ -1,4 +1,5 @@
 import matplotlib.figure
+import pytest
 
 from tests.fixtures import *
 import numpy as np
@@ -122,7 +123,7 @@ def test_predict(self, sklearn_predictor):
         assert isinstance(prediction, float)
         assert 0.0 <= prediction <= 1.0
 
-    def test_plot_shapley_values(self, sklearn_predictor):
+    def test_plot_shapley_values_raises_error(self, sklearn_predictor):
         query = {
             "num_patterns/num_taxa": 0.0,
             "num_sites/num_taxa": 0.0,
@@ -134,5 +135,5 @@ def test_plot_shapley_values(self, sklearn_predictor):
             "proportion_unique_topos_parsimony": 0.0,
         }
 
-        fig = sklearn_predictor.plot_shapley_values(query)
-        assert isinstance(fig, matplotlib.figure.Figure)
+        with pytest.raises(PyPythiaException):
+            sklearn_predictor.plot_shapley_values(query)