📧 SMS Spam Filter using Naive Bayes (MATLAB) 🔍

University Project
Alvaro Bello ✨
Development Date: 14/05/2023 📅

Description 🧠

This project implements a simple spam filter for SMS messages using the Naive Bayes classifier in MATLAB. It classifies messages as either ham (non-spam) or spam based on their word content.

The classifier is trained using the well-known SMSSpamCollection dataset, and uses a Bag of Words model to estimate word frequencies in spam and ham messages, in order to predict if the message is ham or spam.

Project Structure 📁

SMS-Spam-Filter/
├── data/                       # Folder storing datasets
│ └── smsspamcollection/
│ │ └── SMSSpamCollection       # Raw dataset file
│ 
├── extra_functions/            # Provided helper functions (from La Salle University)
├── main.m                      # Main MATLAB script
│ 
├── Project_Documentation.pdf   # Detailed documentation of the project
└── README.md

Results 📈

The script was executed 10 times in total:

• 5 runs with the crossDeleteWords function enabled.
• 5 runs without it.

The experiments yielded an average classification accuracy of 95%, demonstrating not only the effectiveness of the Naive Bayes approach on the SMSSpamCollection dataset, but also that even a simple implementation can achieve highly reliable results in spam detection tasks.

How to Run ▶️

1. Make sure you have MATLAB installed.
2. Download or clone this repository.
3. Run the main script:

>> main.m

Development Environment 🛠️

• Programming Language: Matlab
• IDE: Matlab IDE

Credits 🧑‍🏫

• Dataset: SMSSpamCollection
• Helper functions provided by La Salle University.
• Developed as part of a practical assignment on spam detection and text classification.