fix: ruff rule F841 (unused variable) (#90)

* chore: remove F841 from ruff ignore

* fix violations

* fix: clear out remaining violation

pyproject.toml

@@ -12,7 +12,6 @@ select = ["F"]
 # For now, deselect rules that do not pass
 ignore = [
   "F811", # Redefinition of unused variable
-  "F841", # Local variables assigned but unused
 ]
 
 [tool.ruff.per-file-ignores]

shap/_explanation.py

@@ -776,7 +776,6 @@ def compute_output_dims(values, base_values, data, output_names):
         output_shape = tuple()
 
     interaction_order = len(values_shape) - len(data_shape) - len(output_shape)
-    values_dims = list(range(len(values_shape)))
     output_dims = range(len(data_shape) + interaction_order, len(values_shape))
     return tuple(output_dims)
 
@@ -867,7 +866,6 @@ def _auto_cohorts(shap_values, max_cohorts):
 
     # group instances by their decision paths
     paths = m.decision_path(shap_values.data).toarray()
-    unique_paths = np.unique(m.decision_path(shap_values.data).todense(), axis=0)
     path_names = []
 
     # mark each instance with a path name

shap/explainers/_sampling.py

@@ -54,11 +54,9 @@ def __call__(self, X, y=None, nsamples=2000):
             feature_names = None # we can make self.feature_names from background data eventually if we have it
 
         v = self.shap_values(X, nsamples=nsamples)
-        output_shape = tuple()
         if type(v) is list:
-            output_shape = (len(v),)
             v = np.stack(v, axis=-1) # put outputs at the end
-        e = Explanation(v, self.expected_value, X, feature_names=feature_names)#, output_shape=output_shape)
+        e = Explanation(v, self.expected_value, X, feature_names=feature_names)
         return e
 
     def explain(self, incoming_instance, **kwargs):

shap/plots/_bar.py

@@ -210,7 +210,7 @@ def bar(shap_values, max_display=10, order=Explanation.abs, clustering=None, clu
     ax = pl.gca()
     #xticks = ax.get_xticks()
     bbox = ax.get_window_extent().transformed(fig.dpi_scale_trans.inverted())
-    width, height = bbox.width, bbox.height
+    width = bbox.width
     bbox_to_xscale = xlen/width
 
     for i in range(len(values)):
@@ -233,7 +233,7 @@ def bar(shap_values, max_display=10, order=Explanation.abs, clustering=None, clu
     # put horizontal lines for each feature row
     for i in range(num_features):
         pl.axhline(i+1, color="#888888", lw=0.5, dashes=(1, 5), zorder=-1)
-    
+
     if features is not None:
         features = list(features)
 
@@ -244,7 +244,7 @@ def bar(shap_values, max_display=10, order=Explanation.abs, clustering=None, clu
                     features[i] = int(features[i])
             except:
                 pass # features[i] must not be a number
-    
+
     pl.gca().xaxis.set_ticks_position('bottom')
     pl.gca().yaxis.set_ticks_position('none')
     pl.gca().spines['right'].set_visible(False)
@@ -399,4 +399,4 @@ def bar_legacy(shap_values, features=None, feature_names=None, max_display=None,
     pl.xlabel("SHAP value (impact on model output)")
 
     if show:
-        pl.show()
+        pl.show()

shap/plots/_beeswarm.py

@@ -256,7 +256,6 @@ def beeswarm(shap_values, max_display=10, order=Explanation.abs.mean(0),
 
     # here we build our feature names, accounting for the fact that some features might be merged together
     feature_inds = feature_order[:max_display]
-    y_pos = np.arange(len(feature_inds), 0, -1)
     feature_names_new = []
     for pos,inds in enumerate(orig_inds):
         if len(inds) == 1:
@@ -444,7 +443,6 @@ def summary_legacy(shap_values, features=None, feature_names=None, max_display=N
     # support passing an explanation object
     if str(type(shap_values)).endswith("Explanation'>"):
         shap_exp = shap_values
-        base_value = shap_exp.base_values
         shap_values = shap_exp.values
         if features is None:
             features = shap_exp.data
@@ -701,7 +699,7 @@ def summary_legacy(shap_values, features=None, feature_names=None, max_display=N
                 ds /= np.max(ds) * 3
 
                 values = features[:, i]
-                window_size = max(10, len(values) // 20)
+                # window_size = max(10, len(values) // 20)
                 smooth_values = np.zeros(len(xs) - 1)
                 sort_inds = np.argsort(shaps)
                 trailing_pos = 0
@@ -790,7 +788,7 @@ def summary_legacy(shap_values, features=None, feature_names=None, max_display=N
             # order the feature data so we can apply percentiling
             order = np.argsort(feature)
             # x axis is located at y0 = pos, with pos being there for offset
-            y0 = np.ones(num_x_points) * pos
+            # y0 = np.ones(num_x_points) * pos
             # calculate kdes:
             ys = np.zeros((nbins, num_x_points))
             for i in range(nbins):

shap/plots/_image.py

@@ -81,9 +81,9 @@ def image(shap_values: Explanation or np.ndarray,
         if labels is None:
             labels = shap_exp.output_names
 
-    multi_output = True
+    # multi_output = True
     if not isinstance(shap_values, list):
-        multi_output = False
+        # multi_output = False
         shap_values = [shap_values]
 
     if len(shap_values[0].shape) == 3:

shap/plots/_scatter.py

@@ -84,7 +84,7 @@ def scatter(shap_values, color="#1E88E5", hist=True, axis_color="#333333", cmap=
             ymin = nan_min - (nan_max - nan_min)/20
         if ymax is None:
             ymax = nan_max + (nan_max - nan_min)/20
-        f = pl.subplots(1, len(inds), figsize=(min(6 * len(inds), 15), 5))
+        _ = pl.subplots(1, len(inds), figsize=(min(6 * len(inds), 15), 5))
         for i in inds:
             ax = pl.subplot(1,len(inds),i+1)
             scatter(shap_values[:,i], color=color, show=False, ax=ax, ymin=ymin, ymax=ymax)

shap/plots/_text.py

@@ -263,8 +263,6 @@ def values_min_max(values, base_values):
 
     # build out HTML output one word one at a time
     top_inds = np.argsort(-np.abs(values))[:num_starting_labels]
-    maxv = values.max()
-    minv = values.min()
     out = ""
     # ev_str = str(shap_values.base_values)
     # vsum_str = str(values.sum())

shap/plots/_violin.py

@@ -58,7 +58,6 @@ def violin(shap_values, features=None, feature_names=None, max_display=None, plo
     # support passing an explanation object
     if str(type(shap_values)).endswith("Explanation'>"):
         shap_exp = shap_values
-        base_value = shap_exp.expected_value
         shap_values = shap_exp.values
         if features is None:
             features = shap_exp.data
@@ -98,13 +97,10 @@ def violin(shap_values, features=None, feature_names=None, max_display=None, plo
         else:
             color = colors.blue_rgb
 
-    idx2cat = None
     # convert from a DataFrame or other types
     if str(type(features)) == "<class 'pandas.core.frame.DataFrame'>":
         if feature_names is None:
             feature_names = features.columns
-        # feature index to category flag
-        idx2cat = features.dtypes.astype(str).isin(["object", "category"]).tolist()
         features = features.values
     elif isinstance(features, list):
         if feature_names is None:
@@ -169,7 +165,7 @@ def violin(shap_values, features=None, feature_names=None, max_display=None, plo
                 ds /= np.max(ds) * 3
 
                 values = features[:, i]
-                window_size = max(10, len(values) // 20)
+                # window_size = max(10, len(values) // 20)
                 smooth_values = np.zeros(len(xs) - 1)
                 sort_inds = np.argsort(shaps)
                 trailing_pos = 0
@@ -249,7 +245,7 @@ def violin(shap_values, features=None, feature_names=None, max_display=None, plo
             # order the feature data so we can apply percentiling
             order = np.argsort(feature)
             # x axis is located at y0 = pos, with pos being there for offset
-            y0 = np.ones(num_x_points) * pos
+            # y0 = np.ones(num_x_points) * pos
             # calculate kdes:
             ys = np.zeros((nbins, num_x_points))
             for i in range(nbins):

shap/plots/_waterfall.py

@@ -136,12 +136,10 @@ def waterfall(shap_values, max_display=10, show=True):
             pos_inds.append(0)
             pos_widths.append(-remaining_impact)
             pos_lefts.append(loc + remaining_impact)
-            c = colors.red_rgb
         else:
             neg_inds.append(0)
             neg_widths.append(-remaining_impact)
             neg_lefts.append(loc + remaining_impact)
-            c = colors.blue_rgb
 
     points = pos_lefts + list(np.array(pos_lefts) + np.array(pos_widths)) + neg_lefts + \
         list(np.array(neg_lefts) + np.array(neg_widths))
@@ -161,9 +159,8 @@ def waterfall(shap_values, max_display=10, show=True):
     xlen = plt.xlim()[1] - plt.xlim()[0]
     fig = plt.gcf()
     ax = plt.gca()
-    xticks = ax.get_xticks()
     bbox = ax.get_window_extent().transformed(fig.dpi_scale_trans.inverted())
-    width, height = bbox.width, bbox.height
+    width = bbox.width
     bbox_to_xscale = xlen/width
     hl_scaled = bbox_to_xscale * head_length
     renderer = fig.canvas.get_renderer()
@@ -443,12 +440,10 @@ def waterfall_legacy(expected_value, shap_values=None, features=None, feature_na
             pos_inds.append(0)
             pos_widths.append(-remaining_impact)
             pos_lefts.append(loc + remaining_impact)
-            c = colors.red_rgb
         else:
             neg_inds.append(0)
             neg_widths.append(-remaining_impact)
             neg_lefts.append(loc + remaining_impact)
-            c = colors.blue_rgb
 
     points = pos_lefts + list(np.array(pos_lefts) + np.array(pos_widths)) + neg_lefts + \
         list(np.array(neg_lefts) + np.array(neg_widths))
@@ -468,9 +463,8 @@ def waterfall_legacy(expected_value, shap_values=None, features=None, feature_na
     xlen = plt.xlim()[1] - plt.xlim()[0]
     fig = plt.gcf()
     ax = plt.gca()
-    xticks = ax.get_xticks()
     bbox = ax.get_window_extent().transformed(fig.dpi_scale_trans.inverted())
-    width, height = bbox.width, bbox.height
+    width = bbox.width
     bbox_to_xscale = xlen/width
     hl_scaled = bbox_to_xscale * head_length
     renderer = fig.canvas.get_renderer()

shap/utils/_general.py

@@ -59,10 +59,9 @@ def potential_interactions(shap_values_column, shap_values_matrix):
     index values for SHAP see the interaction_contribs option implemented in XGBoost.
     """
 
-    # ignore inds that are identical to the column 
+    # ignore inds that are identical to the column
     ignore_inds = np.where((shap_values_matrix.values.T - shap_values_column.values).T.std(0) < 1e-8)
-
-    values = shap_values_matrix.values
+
     X = shap_values_matrix.data
 
     if X.shape[0] > 10000: