✅ 06 — Submission Checklist¶
Notebook Checklist¶
- [ ] problem statement
- [ ] data dictionary
- [ ] EDA
- [ ] cleaning decisions
- [ ] feature engineering
- [ ] baseline model
- [ ] final model
- [ ] metrics
- [ ] error analysis
- [ ] business recommendation
Code Checklist¶
- [ ] no hardcoded local-only paths
- [ ] train/test split before preprocessing
- [ ] pipeline used where possible
- [ ] random states set
- [ ] no target leakage
- [ ] final test score reported once
Interview Story¶
Prepare a 2-minute explanation:
I built a churn prediction model. The main business objective was recall for churners. I cleaned missing values, engineered tenure and spend features, compared baseline, logistic regression, and tree models, and selected the model that best balanced recall and precision. The biggest limitation is dataset size and missing customer interaction history.