Verification of dtypes of columns of X_row* is same that self.X

explainerdashboard/explainer_methods.py

@@ -657,6 +657,7 @@ def _permutation_importance(
             if pass_nparray:
                 scores.append(scorer(model, X.values, y.values))
             else:
+                check_dtype_of(col_list, X, old_cols)
                 scores.append(scorer(model, X, y))
 
             X[col_list] = old_cols
@@ -925,14 +926,18 @@ def get_pdp_df(
     for grid_value in grid_values:
         dtemp = X_sample.copy()
         if isinstance(feature, list):
+            tmp_feats = feature.copy()
             if grid_value in X_sample.columns:
                 assert set(X_sample[grid_value].unique()).issubset({0, 1}), (
                     f"{grid_values} When passing a list of features these have to be onehotencoded!"
                     f"But X_sample['{grid_value}'].unique()=={list(set(X_sample[grid_value].unique()))}"
                 )
             dtemp.loc[:, feature] = [1 if col == grid_value else 0 for col in feature]
         else:
+            tmp_feats = [feature]
             dtemp[[feature]] = grid_value
+
+        check_dtype_of(tmp_feats, dtemp, X_sample)
         if is_classifier:
             if cast_to_float32:
                 dtemp = dtemp.values.astype("float32")
@@ -1791,3 +1796,25 @@ def get_xgboost_preds_df(xgbmodel, X_row, pos_label=1):
             0, "pred_proba"
         ]
     return xgboost_preds_df
+
+
+def check_dtype_of(
+    features: List[str],
+    df_target: pd.DataFrame,
+    df_origin: pd.DataFrame
+):
+    """
+        Check if the features in the df_target has the same dtype in df_origin,
+        if not, it fixes the dtype in df_target
+    :param features:
+    :param df_target:
+    :param df_origin:
+    """
+    # Verify if the features has the same dtype as the X sample
+    if (
+            df_target is not None and
+            not df_target[features].dtypes.eq(df_origin[features].dtypes).all()
+    ):
+        df_target[features] = df_target[features].astype(
+            df_origin[features].dtypes
+        )

explainerdashboard/explainers.py

@@ -50,6 +50,7 @@
 
 
 from .explainer_methods import *
+from .explainer_methods import check_dtype_of
 from .explainer_plots import *
 
 
@@ -241,7 +242,9 @@ def __init__(
             col for col in self.regular_cols if not is_numeric_dtype(self.X[col])
         ]
         self.categorical_dict = {
-            col: sorted(self.X[col].unique().tolist()) for col in self.categorical_cols
+            col: sorted(
+                v for v in self.X[col].unique().tolist() if not pd.isna(v)
+            ) for col in self.categorical_cols
         }
         self.cat_cols = self.onehot_cols + self.categorical_cols
         self.original_cols = self.X.columns
@@ -1675,6 +1678,7 @@ def get_contrib_summary_df(
         Returns:
           pd.DataFrame
         """
+        check_dtype_of(self.X.columns, X_row, self.X)
         contrib_df = self.get_contrib_df(index, X_row, topx, cutoff, sort, pos_label)
         return get_contrib_summary_df(
             contrib_df,
@@ -1974,6 +1978,7 @@ def plot_contributions(
 
         """
         assert orientation in ["vertical", "horizontal"]
+        check_dtype_of(self.X.columns, X_row, self.X)
         contrib_df = self.get_contrib_df(
             index=index,
             X_row=X_row,
@@ -2308,6 +2313,7 @@ def plot_pdp(
           plotly.Fig: fig
 
         """
+        check_dtype_of(self.X.columns, X_row, self.X)
         pdp_df = self.pdp_df(
             col,
             index,
@@ -4207,6 +4213,7 @@ def prediction_result_df(self, index=None, X_row=None, round=3):
         if index is not None:
             X_row = self.get_X_row(index)
         if X_row is not None:
+            check_dtype_of(self.X.columns, X_row, self.X)
             if matching_cols(X_row.columns, self.merged_cols):
                 X_row = X_cats_to_X(X_row, self.onehot_dict, self.X.columns)
         if self.shap == "skorch":

requirements.txt

@@ -4,6 +4,7 @@ dash-bootstrap-components>=1
 dash>=2.10.1
 dtreeviz>=2.1
 flask_simplelogin
+flask==3.0.2
 Flask-WTF>=1.1
 graphviz>=0.18.2
 joblib