[SC-9912] Fix index issue when using input_grid

[juanmleng]

Internal Notes for Reviewers

Fix InvalidIndexError: Non-unique indices during DataFrame concatenation. The error occurs because pandas is trying to concatenate DataFrames where the indices are not unique, which happens when we're combining results from multiple features and their bins.

External Release Notes

[github-actions]

PR Summary

This pull request refactors the WeakspotsDiagnosis function in the WeakspotsDiagnosis.py file to improve data handling and metric computation. The key changes include:

1. Renaming Columns: The column previously named "Shape" has been renamed to "Number of Records" to better reflect its content.

2. DataFrame Initialization: The results DataFrames results_1 and results_2 are now initialized outside the loop, allowing for the accumulation of results across iterations.

3. Refactoring Metric Computation: The metric computation logic has been refactored to use temporary dictionaries r1 and r2 for storing intermediate results, which are then concatenated into the main results DataFrames.

4. Improved Sorting and Indexing: The final DataFrame is now reset for its index and sorted by "Feature" and "Dataset" columns, ensuring a consistent and clear output format.

These changes aim to enhance the clarity and efficiency of the code, particularly in how results are accumulated and presented.

Test Suggestions

• Verify that the column name change from 'Shape' to 'Number of Records' is correctly reflected in all outputs.
• Test the WeakspotsDiagnosis function with datasets containing both categorical and continuous features to ensure correct binning and metric computation.
• Check that the final DataFrame output is correctly sorted by 'Feature' and 'Dataset'.
• Ensure that the function handles empty dataframes correctly without errors.
• Validate that the visual assessment logic for weak spots remains functional and accurate.

[cachafla]

Nice