Adapt params for rank stability.

params.yaml

@@ -11,7 +11,7 @@ settings:
     sample_data: false
     calculate_values: false
     calculate_threshold_characteristics: false
-    evaluate_curves: false
+    evaluate_curves: true
     evaluate_metrics: true
     render_plots: true
 
@@ -84,7 +84,7 @@ experiments:
         fn: top_fraction
         alpha_range:
           from: 0.01
-          to: 0.05
+          to: 0.5
           step: 0.01
         plots:
           - rank_stability

src/re_classwise_shapley/curve.py

@@ -48,13 +48,16 @@ def curve_top_fraction(
         alpha_range: A dictionary containing from, to and step keys.
 
     Returns:
-        A pd.Series contianing the alpha value on the x-axis and a unfolded list on the
+        A pd.Series containing the alpha value on the x-axis and a unfolded list on the
             y-axis.
     """
     assert -1 <= alpha_range["to"] <= 1.0
     assert -1 <= alpha_range["from"] <= 1.0
-    n = int((alpha_range["to"] - alpha_range["from"]) / alpha_range["step"]) + 1
-    alpha_range = np.arange(alpha_range["from"], alpha_range["to"], alpha_range["step"])
+    alpha_range = np.arange(
+        alpha_range["from"],
+        alpha_range["to"] + alpha_range["step"],
+        alpha_range["step"],
+    )
     values.sort(reverse=np.all(alpha_range >= 0))
 
     alpha_values = []