📘 Scikit-Learn Intro Study Plan

This repository contains a concise, hands-on study plan to get comfortable applying Scikit-Learn for practical machine learning tasks. The focus is on understanding the core API, building end-to-end workflows, and preparing for code interviews with minimal but effective coverage.

---

🎯 Goals

• Understand the Scikit-Learn API and the fit/transform/predict pattern.
• Build reproducible pipelines for preprocessing and modeling.
• Evaluate models with proper train/test splits and metrics.
• Tune models quickly with cross-validation and simple search.
• Apply to small tabular datasets end-to-end.

---

🧱 Core Concepts to Learn

• Estimators, Transformers, Pipelines
• Train/test split, cross-validation
• Feature preprocessing: scaling, encoding, imputation
• Model evaluation: classification vs regression metrics
• Model selection: GridSearchCV, RandomizedSearchCV
• Saving/loading models (joblib)

---

🛠️ Minimal Toolkit

• Pipeline, ColumnTransformer
• StandardScaler, OneHotEncoder, SimpleImputer
• train_test_split, cross_val_score
• GridSearchCV, RandomizedSearchCV
• Baselines: DummyClassifier, DummyRegressor
• Models: LogisticRegression, RandomForestClassifier, RandomForestRegressor, GradientBoostingRegressor

---

📚 Study Sequence (3–5 hours)

1. Quick API tour with a toy dataset (iris, boston alternative: fetch_california_housing).
2. Preprocessing with ColumnTransformer (numeric vs categorical).
3. Pipelines: preprocessing + model in one object.
4. Evaluation: train_test_split, cross_val_score, metrics (accuracy, f1, roc_auc, rmse/mae).
5. Hyperparameter tuning with GridSearchCV/RandomizedSearchCV.
6. Export model with joblib and reload for inference.

---

🧪 Template: Tabular Classification