Model Information
System Architecture & Classification Type
Model Name
Optimized Random Forest
Class Balancing
Problem Type
Binary Classification
Loan Default Prediction
Target Variable
Loan Default Status
0 = Non-Default, 1 = Default
Training Details
Dataset TypeStructured Tabular
Input Features16 Features
Class BalanceBalanced Weights
PreprocessingStandardScaler
Performance Metrics
Metrics on final validation set
73%Accuracy
62%Recall
24%Precision
0.35F1-Score
CV_MEAN_F1: 0.34
Selection Rationale
- Best balance between recall and precision.
- High stability across cross-validation.
- Effectively identifies high number of defaulters.
- Optimized for imbalanced financial data.
Interpreting Predictions
Low Risk
Low probability of default. Safe to proceed.
High Risk
High probability of default. Caution advised.
* THRESHOLD_CALIBRATED_DURING_TRAINING