add permutation

pages/ALE.py

@@ -66,6 +66,7 @@ def main():
     
     st.title("ALE (Accumulated Local Effects)")
     st.write(prompt_params.ALE_INTRODUCTION)
+    st.write("now has bug, waiting for fix")
     # Explain the selected sample
     if st.button("Explain Sample"):
         explain_example()

pages/PermutationFeatureImportance.py

@@ -50,7 +50,6 @@ def explain_example():
     st.pyplot(fig)
 
     fig, ax = plt.subplots(figsize=(10, 5))
-    st.write("2D Second-Order ALE Plot")
     ax.boxplot(perm_imp.importances[sorted_idx].T,
                vert=False, labels=X_test.columns[sorted_idx])
     ax.set_title("Permutation Importances")
@@ -65,7 +64,7 @@ def main():
         train_model()
     
     st.title("ALE (Accumulated Local Effects)")
-    st.write(prompt_params.ALE_INTRODUCTION)
+    st.write(prompt_params.PERMUTTATION_INTRODUCTION)
     # Explain the selected sample
     if st.button("Explain Sample"):
         explain_example()

src/prompt_config.py

@@ -118,3 +118,15 @@ The process of ALE includes the following steps:
 
 By using ALE, **interpretability** improves by capturing localized effects while mitigating bias from correlated features, making model explanations more reliable.
 """
+PERMUTATION_INTRODUCTION = """
+Permutation Feature Importance is an interpretable machine learning technique that evaluates feature importance by measuring the impact of shuffling a feature’s values on model error.
+
+The process of Permutation Feature Importance includes the following steps:
+1. **Compute Initial Model Error**: Measure the model’s baseline performance using a metric like Mean Squared Error (MSE).
+2. **Permute a Feature**: Randomly shuffle the values of a selected feature while keeping all other features unchanged.
+3. **Compute New Model Error**: Re-evaluate the model on the dataset with the shuffled feature to obtain a new error score.
+4. **Calculate Feature Importance**: Compute the difference between the new error and the original error. A larger difference indicates higher importance.
+5. **Repeat for Stability**: Perform multiple repetitions of permutation and calculate the mean and standard deviation of feature importance scores for reliability.
+
+By using Permutation Feature Importance, **interpretability** improves by providing a direct measure of a feature’s contribution to predictive performance, making model explanations more intuitive and robust.
+"""