This repo now contains an offline-capable ML workflow to mine patterns from:

• Personal Finance Case.pdf
• personal_finance_dataset.xlsx

The pipeline does:

• XLSX ingestion without openpyxl (parses workbook XML directly)
• Feature engineering (debt, liquidity, home equity, ratios)
• Supervised net-worth modeling (PWNETWPG) via ridge regression with CV
• Unsupervised household segmentation via KMeans
• Anomaly detection via Mahalanobis distance
• Export of tables and a markdown summary report

1. (Optional) create and activate a virtual environment:

python3 -m venv .venv
source .venv/bin/activate

2. Install dependencies:

pip install -r requirements.txt

or run:

./scripts/setup_env.sh

3. Run analysis:

python3 analysis/run_finance_ml.py --xlsx personal_finance_dataset.xlsx --outdir outputs --seed 42

The script writes these files to outputs/:

• metrics.json
• analysis_summary.md
• factor_loadings.csv
• cluster_profiles.csv
• anomalies.csv
• engineered_dataset_with_cluster.csv
• dictionary_cleaned.csv
• resilience_scores.csv
• resilience_tier_summary.csv
• resilience_cluster_summary.csv
• resilience_scenario_summary.csv
• resilience_transition_matrix.csv
• resilience_bootstrap_stability.json

• In this environment, internet package download is blocked, so pip install may fail here.
• The analysis code itself is written to run fully offline once required packages are available.