📧 Spam Email Classification using Machine Learning

Hi, I’m Ajay 👋
This repository contains my Spam Email Classification project, where I built and compared multiple ML models to detect whether an email is Spam or Not Spam (Ham).

This project was inspired by the CampusX Spam Classification project, but I extended it with my own feature engineering, model comparison, and performance benchmarking.

---

🔍 Project Overview

Email spam is a huge problem, with billions of unwanted emails sent daily. To tackle this, I:

• Collected and preprocessed a labeled email dataset.
• Cleaned the text (removing punctuation, stopwords, special characters).
• Used TF-IDF vectorization to represent text numerically.
• Trained multiple models including Naive Bayes, Random Forest, Extra Trees, and XGBoost.
• Evaluated them on precision, recall, F1-score, and accuracy.

---

🛠️ Tech Stack

• Python
• Pandas, NumPy – data preprocessing
• Scikit-learn – ML models, TF-IDF vectorization, metrics
• XGBoost – advanced boosting model
• Matplotlib, Seaborn – visualization

---

📊 Model Performance

Model	Accuracy	Precision	Recall	F1-Score
Naive Bayes	94.10%	100%	92%	95.8%
Random Forest	96.8%	97%	96%	96.5%
Extra Trees	97.2%	97%	97%	97%
XGBoost (Final)	98.3%	98%	98%	98%

✨ The XGBoost model was chosen as the final model because it gave the best balance of precision and recall, ensuring fewer false positives and negatives.

---

🌟 Key Insights

• Naive Bayes is simple and surprisingly effective for text classification.
• Ensemble models (Random Forest, Extra Trees, XGBoost) consistently outperform simpler models.
• Precision matters a lot — flagging a real email as spam is more harmful than missing a spam mail.
• Feature engineering (like removing rare words and normalizing text) significantly improves results.

---

🚀 Future Improvements

• Build a Flask/Streamlit app for real-time email classification.
• Deploy the model as an API.
• Add support for multi-language spam detection.
• Train on larger, real-world datasets (Enron, SpamAssassin).

---

📬 Connect with Me

LinkedIn: linkedin.com/in/budida-ajay

GitHub: github.com/ajay9704

---

📌 How to Run

# Clone the repository
git clone https://github.com/ajay9704/spam-classification.git

# Navigate into the project
cd spam-classification

# Install dependencies
pip install -r requirements.txt




# Run the model training
python main.py