How chandrahaas11/minor-project works
This is a command-line explainable-AI pipeline for detecting financial distress risk in the Give-Me-Some-Credit dataset. Its principal workflow loads and prepares ARFF data, explores it, trains and evaluates classifiers, generates SHAP and LIME explanations, and writes a summary report. The main module orchestrates the stages; preprocessing and the modeling and explanation modules provide the core internal computation. Source excerpts show the orchestration and several outputs; some report and explanation details are documented in source comments, while the README is empty.
Main entry points: main.py Pipeline runner · model.py Train and evaluate · explainability.py SHAP explanations · plots Analysis and model plots
Components 13
Pipeline orchestration
- Pipeline runner
project/src/main.py
Data preparation and analysis
- Preprocessing
project/src/preprocessing.py - Exploratory analysis
project/src/eda.py
Risk modeling
- Train and evaluate
project/src/model.py
Model explanations
- SHAP explanations
project/src/explainability.py - LIME explanations
project/src/explainability.py
Results
- Report generation
project/src/report.py - Model metrics
project/results/model_metrics.csv - Analysis and model plots
project/results/plots - Explanation artifacts
project/results/explanations - Pipeline report
project/results/report.md
Other components
- CLI operator
- ARFF dataset
Connections 19
- CLI operator → Pipeline runner: runs
project/src/main.py - ARFF dataset → Preprocessing: loads
project/src/preprocessing.py - Pipeline runner → Preprocessing: calls
project/src/main.py - Preprocessing → Pipeline runner: returns splits
project/src/main.py - Pipeline runner → Exploratory analysis: calls
project/src/main.py - Pipeline runner → Train and evaluate: calls
project/src/main.py - Train and evaluate → Model metrics: writes
project/src/model.py - Train and evaluate → Analysis and model plots: generates
project/src/model.py - Exploratory analysis → Analysis and model plots: saves plots
project/src/eda.py - Pipeline runner → SHAP explanations: calls
project/src/main.py - Pipeline runner → LIME explanations: calls
project/src/main.py - SHAP explanations → Train and evaluate: explains predictions
project/src/explainability.py - LIME explanations → Train and evaluate: explains predictions
project/src/explainability.py - SHAP explanations → Analysis and model plots: saves plots
project/src/explainability.py - SHAP explanations → Explanation artifacts: writes SHAP values
project/src/explainability.py - LIME explanations → Analysis and model plots: saves plots
project/src/explainability.py - LIME explanations → Explanation artifacts: writes explanations
project/src/explainability.py - Pipeline runner → Report generation: calls
project/src/main.py - Report generation → Pipeline report: writes
project/src/report.py