🧠 ML Foundations: Theory to Practice

This repository explores core machine learning concepts through interactive, hands-on Jupyter notebooks. From building a neural network from scratch to applying dimensionality reduction and classification with real-world libraries, this project bridges the gap between theory and practical application.

---

📚 Notebooks Overview

Notebook	Focus	Description
01_pca_dimensionality.ipynb	📉 Dimensionality Reduction	Uses Principal Component Analysis (PCA) to reduce high-dimensional data while preserving variance. Ideal for feature compression and visualization.
02_neural_net_numpy.ipynb	🤖 Neural Network from Scratch	Implements a simple feedforward neural network using only NumPy. Covers forward propagation, backpropagation, and a training loop.
03_classification_models.ipynb	🧪 ML Classification	Applies supervised learning techniques using scikit-learn. Includes data preprocessing, model training, evaluation, and comparison.

---

🔍 Key Takeaways

🎯 PCA

• Reduced a 4D dataset to 2D while retaining ~95% of variance.
• Visualized separability of classes in reduced space.

🛠️ Custom Neural Network

• Built from scratch using NumPy (no ML frameworks).
• Trained a neural net using manual backpropagation and gradient descent.
• Demonstrated convergence on a synthetic dataset.

📈 Classification Pipeline

• Achieved ~97% accuracy using logistic regression.
• Visualized results via a confusion matrix.
• Compared multiple models including SVM and Decision Trees.

---

🧰 Tech Stack

• Python – Core scripting language
• NumPy – Neural network implementation
• Scikit-learn – PCA, classification, evaluation tools
• Matplotlib & Seaborn – Visualization
• Jupyter Notebooks – Interactive development

---

📁 Directory Structure

ml-theory-to-practice/
├── 01_pca_dimensionality.ipynb        # Dimensionality Reduction with PCA
├── 02_neural_net_numpy.ipynb          # Neural Network built from scratch
├── 03_classification_models.ipynb     # Supervised ML classification models
└── README.md

---

🚀 Project Goals

• Solidify foundational machine learning theory through hands-on implementation
• Develop intuition for model behavior, training dynamics, and evaluation
• Showcase practical skills for portfolio use or collaborative work