📊 Machine Learning & Data Science Notebooks

A curated collection of production-quality Jupyter Notebooks covering Exploratory Data Analysis, Classical ML, and Deep Learning with PyTorch and Scikit-Learn.

View Notebooks · Quick Start · Results · Contributing

---

📋 Table of Contents

• About
• Notebook Catalog
• Datasets
• Model Architectures
• Results & Benchmarks
• Quick Start
• Project Structure
• Tech Stack
• Contributing
• License

---

🔍 About

This repository is a hands-on resource for anyone learning Machine Learning, Deep Learning, and Data Science. Each notebook is self-contained, heavily commented, and follows best practices for reproducible research. Whether you are a student exploring neural networks for the first time or a practitioner looking for clean reference implementations, these notebooks provide clear, end-to-end workflows from data loading to model evaluation.

---

📓 Notebook Catalog

#	Notebook	Topic	Difficulty	Framework
01	01_Exploratory_Data_Analysis.ipynb	EDA, feature engineering, missing value imputation, distribution analysis	🟢 Beginner	Pandas, Seaborn, Matplotlib
02	02_Random_Forest_Classifier.ipynb	Classification pipeline with cross-validation, hyperparameter tuning (GridSearchCV), feature importance	🟡 Intermediate	Scikit-Learn
03	03_PyTorch_CNN_MNIST.ipynb	Convolutional Neural Network from scratch, training loop, GPU acceleration, confusion matrix	🔴 Advanced	PyTorch

Tip: Each notebook includes inline explanations, mathematical intuition, and visualization of intermediate results.

---

📊 Datasets Used

Dataset	Source	Size	Task	Notebook
Titanic Survival	Kaggle	891 rows, 12 features	Binary Classification / EDA	01, 02
MNIST Handwritten Digits	Yann LeCun	70,000 images (28x28)	Multi-class Classification	03

---

🧠 Model Architectures

Random Forest Classifier (Notebook 02)

• Ensemble Method: Bagging with 200 decision trees
• Tuning: GridSearchCV over max_depth, n_estimators, min_samples_split
• Evaluation: 5-fold stratified cross-validation, ROC-AUC, precision-recall curves

PyTorch CNN (Notebook 03)

Input (1x28x28)
  → Conv2d(1, 32, 3x3) → BatchNorm → ReLU → MaxPool(2x2)
  → Conv2d(32, 64, 3x3) → BatchNorm → ReLU → MaxPool(2x2)
  → Flatten
  → Linear(1600, 128) → ReLU → Dropout(0.5)
  → Linear(128, 10) → Softmax

• Optimizer: Adam (lr=0.001)
• Loss: CrossEntropyLoss
• Epochs: 15
• Batch Size: 64

---

📈 Results & Benchmarks

Model	Dataset	Accuracy	Precision	Recall	F1 Score
Random Forest (tuned)	Titanic	84.2%	82.1%	79.5%	80.8%
PyTorch CNN	MNIST Test Set	99.1%	99.0%	99.1%	99.0%

Results are reproducible with the seeds set in each notebook. Your results may vary slightly depending on hardware.

---

🚀 Quick Start

Prerequisites

• Python 3.10 or higher
• pip or conda package manager
• (Optional) NVIDIA GPU with CUDA for accelerated training

Option 1: Local Setup

# Clone the repository
git clone https://github.com/razinahmed/machine-learning-notebooks.git
cd machine-learning-notebooks

# Create and activate a virtual environment
python -m venv venv
source venv/bin/activate        # macOS/Linux
# venv\Scripts\activate         # Windows

# Install all dependencies
pip install -r requirements.txt

# Launch Jupyter Notebook
jupyter notebook

Option 2: Google Colab (No Setup Required)

Don't have a local GPU? Open any notebook directly in Google Colab:

Notebook	Colab Link
01 - EDA
02 - Random Forest
03 - PyTorch CNN

Option 3: Conda Environment

conda create -n ml-notebooks python=3.10
conda activate ml-notebooks
pip install -r requirements.txt
jupyter notebook

---

📁 Project Structure

machine-learning-notebooks/
├── 01_Exploratory_Data_Analysis.ipynb    # EDA with Pandas & Seaborn
├── 02_Random_Forest_Classifier.ipynb     # Scikit-Learn ML pipeline
├── 03_PyTorch_CNN_MNIST.ipynb            # Deep Learning with PyTorch
├── data/                                 # Raw and processed datasets
│   ├── titanic_train.csv
│   └── titanic_test.csv
├── models/                               # Saved model checkpoints
├── figures/                              # Generated plots and charts
├── requirements.txt                      # Python dependencies
├── LICENSE
└── README.md

---

🛠️ Tech Stack

Category	Technologies
Languages	Python 3.10+
Data Analysis	Pandas, NumPy, Matplotlib, Seaborn
Machine Learning	Scikit-Learn, XGBoost
Deep Learning	PyTorch, torchvision
Environment	Jupyter Notebook, Google Colab
Utilities	tqdm, joblib, pickle

---

🤝 Contributing

Contributions are welcome! If you have a notebook that demonstrates a useful ML technique:

1. Fork this repository
2. Create a feature branch (git checkout -b add-new-notebook)
3. Add your notebook following the naming convention XX_Notebook_Name.ipynb
4. Ensure all cells run top-to-bottom without errors
5. Submit a Pull Request with a description of the notebook and results

Please make sure your notebooks include:

• Clear markdown explanations between code cells
• Reproducible results with random seeds
• Properly labeled visualizations

---

📄 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

Built with passion for Data Science by Razin Ahmed

If these notebooks helped you learn, please consider giving the repo a ⭐