Machine Learning Module 1:

Overview This repository contains 5 complete machine learning projects implemented from scratch. Each project demonstrates core ML concepts with working code, visualizations, and clear explanations.

Projects Included

Linear Regression from Scratch → Implementation using only NumPy Cost Function Visualization → 3D “bowl-shaped” cost surface Gradient Descent Implementation → Optimization with learning rates Supervised vs Unsupervised → Side-by-side comparison California Housing → Real-world ML pipeline

Quick Start

Option 1: Run Locally (Recommended)

Step 1: Install Python Download Python 3.8 or higher from python.org Make sure “Add Python to PATH” is enabled

Step 2: Download Repository Using Git: git clone https://github.com/YOUR_USERNAME/YOUR_REPO_NAME.git

cd YOUR_REPO_NAME

Or download ZIP and extract

Step 3: Install Dependencies pip install numpy matplotlib pandas seaborn scikit-learn ipywidgets jupyter

Step 4: Launch Jupyter jupyter notebook

Step 5: Run Projects Open any .ipynb file Run cells using Shift + Enter Start with 00_INDEX.ipynb

Option 2: Run on Google Colab (No Installation)

Open notebook on GitHub Click “Open in Colab” OR go to colab.research.google.com Upload the notebook Run cells with Shift + Enter

Enable widgets in Colab: !pip install ipywidgets from google.colab import output output.enable_custom_widget_manager()

Repository Structure

ml-module-1/

00_INDEX.ipynb

project_1_linear_regression_scratch.ipynb project_2_cost_function_visualization.ipynb project_3_gradient_descent_implementation.ipynb project_4_supervised_vs_unsupervised.ipynb project_5_california_housing_real_data.ipynb

README.txt requirements.txt

images/ project1_fit.png project2_3d_surface.png ...

Setup Instructions

Windows pip install numpy matplotlib pandas seaborn scikit-learn ipywidgets jupyter cd C:\path\to\your\ml-module-1 jupyter notebook

Mac brew install python pip3 install numpy matplotlib pandas seaborn scikit-learn ipywidgets jupyter cd /path/to/your/ml-module-1 jupyter notebook

Linux sudo apt install python3-pip pip3 install numpy matplotlib pandas seaborn scikit-learn ipywidgets jupyter cd /path/to/your/ml-module-1 jupyter notebook

Testing Each Project

Project 1: Linear Regression Expected: scatter plot, regression line, cost decreasing graph Result: weights close to [5, 2.5] Success: R² > 0.9

Project 2: Cost Function Visualization Expected: 3D plot, contour plot, sliders Success: plots render correctly

Project 3: Gradient Descent Expected: cost decreases, learning rate comparison Success: stable convergence

Project 4: Supervised vs Unsupervised Expected: regression, clustering, classification visuals Success: clear comparison plots

Project 5: California Housing Expected: stats, heatmap, predictions vs actual Success: R² around 0.6

Troubleshooting

Error: numpy not found → run pip install numpy Error: sklearn not found → run pip install scikit-learn Jupyter not starting → pip install --upgrade jupyter Sliders not working → install ipywidgets and restart Dataset not loading → check internet connection

Colab Fix: !pip install ipywidgets from google.colab import output output.enable_custom_widget_manager()

Environment Test

import sys print(sys.version)

try: import numpy print("NumPy OK") except: print("Install NumPy")

try: import matplotlib print("Matplotlib OK") except: print("Install Matplotlib")

try: import sklearn print("scikit-learn OK") except: print("Install scikit-learn")

FAQ

Do I need to run all projects? No, each project is independent

How long does it take? 1–2 minutes per project

Is my data saved? No, everything runs locally

No installation option? Use Google Colab

Why duplicate classes? Each notebook is self-contained

Performance

Project 1 → ~30 sec, ~100 MB, no internet Project 2 → ~30 sec, ~150 MB, no internet Project 3 → ~30 sec, ~100 MB, no internet Project 4 → ~20 sec, ~100 MB, no internet Project 5 → ~60 sec, ~200 MB, internet required (first run)

Support

Check troubleshooting Ensure Python 3.8+ Install all dependencies Open GitHub issue if needed

License MIT License

Star the repo if you found it useful.

Built from scratch to demonstrate strong understanding of machine learning fundamentals.