Update automative_xai.py

[pranjaykumar926]

No description provided.

[pranjaykumar926]

Modifications Made in Automotive_XAI Class:

1. Added Return Values

   • Each method now returns relevant outputs instead of just printing results.
   • Example:
     • do_PFI() returns sorted_importances (list of feature importances).
     • do_IFI() returns ifi_values.
     • do_LIME() returns the LIME explanation object.
     • do_SHAP() returns shap_values, X_sample, and X_background.
     • do_ALE() returns the generated ALE plot.

2. Fixed LIME Indexing Issue

   • Previously, set_local_index(self.y) attempted to use y directly as an index, which could cause errors.
   • Now, a valid sample index (sample_index = 0) is selected explicitly.

3. Ensured Proper Handling of feature_name

   • In do_ALE() and do_SHAP(), feature_name is now correctly passed when generating plots.

4. Improved Docstrings and Method Descriptions

   • Each method now includes a detailed docstring explaining its purpose, how it works, and what it returns.
   • Example:

     def do_SHAP(self):
         """
         Compute and visualize SHAP (SHapley Additive exPlanations) values.
         SHAP explains model predictions by assigning contributions to each feature.
     
         Returns:
         - shap_values: Computed SHAP values.
         - X_sample: Sample input data.
         - X_background: Background dataset for SHAP calculations.
         """

5. Ensured Consistent Feature Input Handling

   • Used self.X[self.features] uniformly across all methods to avoid input inconsistencies.

Benefits of These Changes:

✅ More modular and reusable code
✅ Better debugging and analysis with return values
✅ Clearer documentation for easier understanding
✅ Avoids potential indexing errors in LIME