avoid collisions in logistic_regression_objective tmp column name

[eordentlich]

No description provided.

[greptile-apps]

Greptile Overview

Greptile Summary

This PR fixes a potential column name collision issue in the logistic_regression_objective function. Previously, the function hardcoded "scaledFeatures" as the output column name for the StandardScaler, which could collide with existing columns in the input DataFrame.

Changes:

• Dynamically generates a unique temporary column name (_objective_tmp) by checking against existing column names
• Uses a while loop to prepend underscores until a unique name is found
• Follows the same collision avoidance pattern already used in params.py (line 118-119)

Impact:

• Prevents runtime errors when input DataFrames contain a column named "scaledFeatures"
• Maintains backward compatibility as this is a utility function used in testing
• No changes to public API or behavior

Confidence Score: 5/5

• This PR is safe to merge with no risk - it's a defensive bug fix with no breaking changes
• The change is a straightforward defensive fix that prevents potential column name collisions. The implementation follows an existing pattern in the codebase (params.py), is well-contained, and only affects a utility function used for testing. No edge cases or issues found.
• No files require special attention

Important Files Changed

File Analysis

Filename	Score	Overview
python/src/spark_rapids_ml/metrics/utils.py	5/5	Adds collision avoidance for temporary column name used by StandardScaler; follows existing codebase patterns

Sequence Diagram

sequenceDiagram
    participant Caller
    participant logistic_regression_objective
    participant DataFrame
    participant StandardScaler
    
    Caller->>logistic_regression_objective: Call with df and lr_model
    logistic_regression_objective->>logistic_regression_objective: Transform df with model
    logistic_regression_objective->>logistic_regression_objective: Calculate log_loss
    
    alt Standardization is True
        logistic_regression_objective->>DataFrame: Get df.columns
        DataFrame-->>logistic_regression_objective: Return column_names
        logistic_regression_objective->>logistic_regression_objective: Initialize outputCol = "_objective_tmp"
        
        loop While outputCol in column_names
            logistic_regression_objective->>logistic_regression_objective: outputCol = "_" + outputCol
        end
        
        logistic_regression_objective->>StandardScaler: Create with inputCol and unique outputCol
        logistic_regression_objective->>StandardScaler: Fit on df
        StandardScaler-->>logistic_regression_objective: Return scaler_model with std
        logistic_regression_objective->>logistic_regression_objective: Adjust coefficients by stdev
    end
    
    logistic_regression_objective->>logistic_regression_objective: Calculate full_objective
    logistic_regression_objective-->>Caller: Return full_objective

Loading

[eordentlich]

build