Hands-On Machine Learning Journey 🚀

Welcome to my Machine Learning repository! This repo serves as my personal codebase, template center, and notebook journal as I master the concepts from the famous book "Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow" by Aurélien Géron.

The goal of this repository is to avoid reinventing the wheel—I document the implementation of pipelines, algorithms, and workflows chapter by chapter so that I can easily reuse these codebases as boilerplate templates for future real-world data science projects.

---

🛠️ Repository Structure & Chapters Covered

📁 Chapter 1: The Machine Learning Landscape

• Concepts: Introduction to ML types (Supervised vs Unsupervised, Batch vs Online, Instance-based vs Model-based), main challenges of ML (insufficient data, non-representative data, overfitting, underfitting).
• Code Implementation: Simple linear model training to predict life satisfaction based on GDP per capita.

📁 Chapter 2: End-to-End Machine Learning Project

• Concepts: Frame the problem, look at the big picture, handle missing data, manage text/categorical attributes, feature scaling, and evaluate performance using Root Mean Squared Error (RMSE).
• Code Templates Built:
  • Stratified Shuffle Split (for unbiased train/test partitioning)
  • Custom Data Preprocessing Pipelines (Pipeline, ColumnTransformer)
  • Handling Missing Values (SimpleImputer) & Text Encoding (OneHotEncoder)
  • Model Selection & Baseline Training (LinearRegression, DecisionTreeRegressor, RandomForestRegressor)

📁 Chapter 3: Classification

• Concepts: Moving away from continuous regression to categorical classification using the MNIST digit dataset. Understanding the deep mechanics of model evaluation.
• Code Templates Built:
  • Stochastic Gradient Descent Classifier (SGDClassifier)
  • Confusion Matrix, Precision, Recall, and F1-Score analysis
  • The Precision/Recall Tradeoff visualization using decision_function
  • Receiver Operating Characteristic (ROC) Curve and Area Under Curve (AUC) score computation
  • Multiclass & Multilabel Classification setups

---

💡 How I Use This Repo (The Smart Workflow)

Rather than memorizing Scikit-Learn syntax, this repository acts as my custom reference manual. Whenever I face a new dataset:

1. I open my corresponding chapter folder or baseline template.
2. I copy the tested pipeline structural code.
3. I replace the dataset source and tweak hyper-parameters.
4. Focus remains on Logic, Strategy, and Tuning, not syntax!

---

👨‍💻 Developer Credit

• Developed and Maintained by: Vinay Kumar Mandal
• Language & Stack: Python, Jupyter Notebooks, Scikit-Learn, Pandas, NumPy.

---

Feel free to explore the notebooks! Every major block contains inline comments explaining the "Why" behind the code, matching my physical book annotations.