How mainak9093/company_bankruptcy_modelling works
This repository is a reproducible bankruptcy-risk modeling project, not a hosted service. Its main workflow loads labeled financial data, fits a leakage-safe preprocessing and modeling pipeline, tunes a decision threshold on validation data, and evaluates on held-out data. Separate entry points provide cross-validation, baseline comparisons, inference, labeled-data scoring, and report-figure generation. The graph follows documented workflows and shown imports/calls; some pipeline internals and report-generation details are only partially excerpted.
Main entry points: pipeline.py Fold pipeline · baselines.py Baseline comparison · predict.py Prediction · train.py Holdout training
Components 17
Experiment workflows
- Holdout training
src/train.py - Cross-validation
src/cross_validate.py - Baseline comparison
src/baselines.py
Model pipeline
- Fold pipeline
src/pipeline.py - Data cleaning
src/data_preprocessing.py - ANOVA selection
src/feature_selection.py - Model training
src/model_training.py - Metrics and threshold
src/model_evaluation.py - Settings
src/config.py
Scoring and artifacts
- Saved artifacts
models - Prediction
src/predict.py - Labeled scoring
src/evaluate.py
Reporting
- Report figures
src/make_report_figures.py - Experiment results
results
Other components
- Analyst
- Labeled data
- Scoring CSV
Connections 36
- Analyst → Holdout training: runs
README.md - Labeled data → Holdout training: loads
README.md - Holdout training → Fold pipeline: builds fold
src/train.py - Holdout training → Metrics and threshold: formats metrics
src/train.py - Holdout training → Saved artifacts: saves models
src/train.py - Holdout training → Experiment results: writes holdout
src/train.py - Analyst → Cross-validation: runs
README.md - Labeled data → Cross-validation: loads
src/cross_validate.py - Cross-validation → Fold pipeline: builds folds
src/cross_validate.py - Cross-validation → Experiment results: writes summaries
src/cross_validate.py - Analyst → Baseline comparison: runs
README.md - Labeled data → Baseline comparison: loads
src/baselines.py - Baseline comparison → Data cleaning: fits cleaner
src/baselines.py - Baseline comparison → ANOVA selection: selects features
src/baselines.py - Baseline comparison → Model training: fits baselines
src/baselines.py - Baseline comparison → Metrics and threshold: tunes and scores
src/baselines.py - Baseline comparison → Experiment results: writes comparison
src/baselines.py - Fold pipeline → Data cleaning: cleans splits
src/pipeline.py - Fold pipeline → ANOVA selection: selects features
src/pipeline.py - Fold pipeline → Model training: scales and fits
src/pipeline.py - Fold pipeline → Metrics and threshold: tunes and evaluates
src/pipeline.py - Fold pipeline → Settings: reads settings
src/pipeline.py - Analyst → Prediction: runs
README.md - Scoring CSV → Prediction: loads
README.md - Prediction → Saved artifacts: loads artifacts
src/predict.py - Prediction → Data cleaning: transforms inputs
src/predict.py - Prediction → ANOVA selection: selects columns
src/predict.py - Prediction → Model training: gets probabilities
src/predict.py - Analyst → Labeled scoring: runs
README.md - Scoring CSV → Labeled scoring: loads labeled CSV
src/evaluate.py - Labeled scoring → Prediction: prepares features
src/evaluate.py - Labeled scoring → Model training: gets probabilities
src/evaluate.py - Labeled scoring → Metrics and threshold: computes metrics
src/evaluate.py - Analyst → Report figures: runs
README.md - Labeled data → Report figures: loads
src/make_report_figures.py - Report figures → Fold pipeline: builds fold
src/make_report_figures.py