00_Preamble

Table of Contents

• The ML Project Life Cycle
• 1. EDA
• 2. Data Cleaning
• 3. Data Imputation
• 4. Feature Engineering & Encoding
• 5. Feature Scaling
• 6. Train-Test Split & Cross-Validation
• 7. Hyperparameter Tuning
• 8. Evaluation Metrics — Regression
• 9. Evaluation Metrics — Classification
  • Confusion Matrix
  • ROC-AUC Curve
  • Precision-Recall Curve
• 10. scikit-learn Models
  • Linear Regression
  • Ridge & Lasso
  • Logistic Regression
  • SVM
  • Decision Trees
  • Ensemble Methods
  • KNN
  • Naïve Bayes
  • Neural Networks
  • K-Means
  • DBSCAN
  • PCA
• 11. scikit-learn Pipeline API
• 12. Inference & Deployment Checklist
• Appendix — Legacy Notes

This file is a comprehensive reference for the IITM BSCS2008 Machine Learning Practice course, covering the full end-to-end project lifecycle in scikit-learn. Use it as your single-stop interview guide.

---

The ML Project Life Cycle

Every ML project follows a repeatable sequence of steps. Think of it as a pipeline where a mistake early on propagates and amplifies at every downstream stage.

Raw Data
   ↓
EDA (understand the data)
   ↓
Train–Validation–Test Split
   ↓
Data Cleaning (remove noise and inconsistencies)
   ↓
Data Imputation (handle missing values)
   ↓
Feature Engineering & Encoding
   ↓
Feature Scaling / Normalization
   ↓
Model Selection & Training
   ↓
Hyperparameter Tuning (with Cross-Validation)
   ↓
Evaluation on held-out Test Set
   ↓
Inference / Deployment

---