📩 Spam Message Detection (ML-Based)

A machine learning project that classifies SMS messages as spam or ham using text preprocessing, TF-IDF feature extraction, and a Multinomial Naive Bayes classifier.
This project also includes an interactive spam checker that allows users to classify custom messages in real-time through the terminal.

---

📌 Table of Contents

• Overview
• Features
• Dataset
• Model Details
• Project Structure
• Installation
• Usage
• Evaluation
• Future Enhancements
• Requirements
• License
• Author

---

📖 Overview

This project trains a machine learning model to detect whether a message is spam (unwanted advertisements, fraud messages) or ham (legitimate messages).
It uses:

• Regex-based text cleaning
• TF-IDF vectorizer
• Multinomial Naive Bayes
• Train/Test splitting with stratification
• Detailed evaluation metrics

The script also includes an interactive input mode to test custom messages.

---

✨ Features

• Clean & preprocess raw SMS messages
• Text vectorization using TF-IDF
• Highly accurate Multinomial Naive Bayes classifier
• Interactive spam detection via terminal
• 98.6% test accuracy
• Full classification report
• Easily extendable for real-time filtering

---

📊 Dataset

Dataset used:

SMS Spam Collection Dataset
Source (Kaggle):
https://www.kaggle.com/datasets/uciml/sms-spam-collection-dataset

Dataset statistics:

• 5572 messages total
• 747 spam
• 4825 ham

Columns used:

• label — spam or ham
• text — SMS message content

---

🧠 Model Details

🔧 Pipeline

Pipeline([

('tfidf', TfidfVectorizer(ngram_range=(1, 2), min_df=2)),

('nb', MultinomialNB(alpha=0.1))

])

🧹 Text Cleaning

Performed using the clean_text() function:

• Lowercasing
• Removing URLs
• Removing email addresses
• Removing special characters
• Normalizing whitespace

🔍 Train/Test Split

• 80% training
• 20% testing
• Stratified
• random_state=42

---

📁 Project Structure

spam-message-detection/

├── spam_message.py # Main ML script + interactive checker

├── spam.csv # Dataset

├── Requirements.txt # Listed dependencies

├── README.md # Project documentation

└── .gitignore

---

🚀 Installation

1. Clone the repository

   git clone https://github.com/suryat8957/spam-message-detection.git
   cd spam-message-detection
   

2. Install required packages

   pip install -r Requirements.txt
   

3. Run the program

   python spam_message.py
   

---

▶️ Usage

Interactive Spam Checker

When running the script: --- Interactive Spam Checker --- Type a message or 'quit' to exit

Enter message > Free entry in a prize draw!! => SPAM (spam probability: 0.9823)

Predict inside Python

label, prob = predict_message("Congratulations! You have won a reward") print(label, prob)

---

📈 Evaluation

Performance on test data:

Metric	Score
Accuracy	0.986
Precision (Spam)	0.99
Recall (Spam)	0.90
F1-Score (Spam)	0.94
Weighted Avg F1	0.99

Summary

• Ham detection: ~100% accuracy
• Spam detection: 90% recall
• Very strong overall performance

---

🔮 Future Enhancements

Planned additions:

• Real-time SMS/email/WhatsApp filtering
• Deploy as a web API (FastAPI / Flask)
• Web UI for message testing
• More ML models (Logistic Regression, SVM, BERT)
• Export trained model for production use

---

📦 Requirements

As listed in Requirements.txt (example):

pandas scikit-learn

(Actual versions depend on your file.)

---

📜 License

This project is open-source.
You may add a license file such as MIT if needed.

---

👤 Author

Surya
GitHub: https://github.com/suryat8957