feat(ml): train fix success predictor #180

[prasiddhi-105]

Linked issue

Closes #180

What this PR does

Adds a training script (backend/scripts/train_fix_predictor.py) that fits a Logistic Regression model to predict whether proposed code fixes will pass verification. It reads telemetry data from SQLite or CSV, one-hot encodes features, evaluates performance (accuracy and ROC-AUC) on a 20% test split, and serializes the resulting model to backend/app/ml/models/fix_predictor.pkl.

Type of change

• Bug fix
• New feature
• ML model / training pipeline
• Refactor (no behaviour change)
• Documentation
• Tests only

ML tier (if applicable)

• Tier 1 — Triage
• Tier 2 — Predictive
• Tier 3 — Autonomous
• Not ML-related

Stack affected

• Backend
• Frontend
• Both

---

Changes

Backend

• Created backend/scripts/train_fix_predictor.py:
  • Supports loading telemetry data from SQLite DB or CSV.
  • Enforces minimum sample size (≥ 100 rows) requirement and exits gracefully if data is insufficient.
  • One-hot encodes categorical variables and splits data into an 80/20 train/test split.
  • Fits a LogisticRegression model and outputs accuracy and ROC-AUC scores to stdout.
  • Saves the trained model to backend/app/ml/models/fix_predictor.pkl.
  • Added CLI argument parsing including --help flag support.
• Updated .gitignore to ensure .pkl model files are ignored.

Frontend

• No frontend changes.

New dependencies

• No new dependencies added (scikit-learn, pandas, and joblib are already present).

Database / schema changes

• No schema changes.

---

Testing

How did you test this?

• Tested --help flag: python backend/scripts/train_fix_predictor.py --help.
• Verified script execution when dataset < 100 samples (correctly prints requirement message and exits).
• Confirmed fix_predictor.pkl is excluded by .gitignore.

Checklist

• Tested locally end-to-end
• New ML model falls back gracefully when model file is absent
• No new console.error or unhandled Python exceptions introduced
• Added or updated tests where applicable
• requirements.txt / package.json updated if new dependencies added
• New model files (.pkl, .pt, etc.) are gitignored, not committed

---

Anything reviewers should focus on

Please check feature encoding and metric calculations in train_fix_predictor.py.

Screenshots (if UI changed)

N/A (CLI training script)

[github-actions]

🎉 Thank you @prasiddhi-105 for submitting a Pull Request!

We're excited to review your contribution.

✅ Before Review

• Ensure all CI checks pass.
• Complete the PR template.
• Link the related issue.

⚡ Want faster reviews and contributor support?

Join our Discord community:

🔗 https://discord.gg/FcXuyw2Rs

Maintainers and mentors are active there and can help resolve blockers quickly.

Happy Contributing! 🚀