📩 SMS / Email Spam Classifier

This project is a machine learning–powered web application that classifies SMS or email messages as Spam or Not Spam. It uses natural language processing (NLP) techniques with TF-IDF vectorization and classical machine-learning models, deployed as an interactive web app using Streamlit.

---

🚀 Live Demo

👉[] https://sms-spam-detection-by-shraddha.streamlit.app/

---

Dataset - SMS Spam Collection Dataset

👉 https://www.kaggle.com/datasets/uciml/sms-spam-collection-dataset

📌 Features

• Text preprocessing with NLTK
• TF-IDF vectorization
• Multiple models evaluated (Naive Bayes, SVM, ensembles, etc.)
• Final trained model used is naive bayes
• Interactive Streamlit interface
• Cloud-deployable via GitHub

---

🧠 Machine Learning Pipeline

1. Text cleaning & normalization
2. Tokenization and stopword removal
3. TF-IDF feature extraction
4. Model training & evaluation
5. Model serialization using Pickle
6. Deployment using Streamlit

---

📂 Project Structure

sms-spam-detection/
│
├── app.py # Streamlit app
├── model.pkl # Trained ML pipeline
├── vectorizer.pkl
├── requirements.txt # Dependencies
├── Sms_Spam_classifier.ipynb # Notebook
├── README.md # Project documentation
├── .gitignore
├── screenshots/ # App screenshots for README

🛡️ NLTK Resources

The app automatically downloads required NLTK resources (punkt, punkt_tab, and stopwords) at runtime so that tokenization works correctly in cloud deployments.

📊 Models Evaluated

Several models were trained and compared:

• Multinomial Naive Bayes

• Support Vector Machine

• Logistic Regression

• Random Forest

• Gradient Boosting

• Voting & Stacking Ensembles

The best-performing model was selected and deployed.

---

⚙️ Installation & Running Locally

Clone the repository

git clone https://github.com/smshelar/sms-spam-detection.git
cd sms-spam-detection

🧪 Example Inputs

Spam
Congratulations! You have won a free voucher. Click now!

Not Spam
Hi, are we still meeting tomorrow afternoon?

🔮 Future Improvements

• Deep-learning–based text models

• Language detection

• Probability score display

• Model explainability (SHAP / feature importance)

• Dataset expansion

👩‍💻 Author

Shraddha Shelar