SMS Spam Classification

This project is a web service for SMS spam classification, built using the Flask framework and deployed on Render. It leverages a trained machine learning model to classify SMS messages as either HAM (legitimate) or SPAM.

Overview

This web service provides a user interface and an API endpoint to classify SMS messages as spam or ham based on a trained machine learning model. The application utilizes a Random Forest Classifier and TF-IDF vectorizer, stored in rfc-model.pkl and tfidf-vectorizer.pkl files, respectively.

Setup

Prerequisites

• Python 3.8+

• nltk

• scikit-learn

• Flask

  For more details, refer to the requirements.txt for this project.

Installation

1. Clone the repository :

   git clone https://github.com/nirmit27/SMS-Spam-Classifier.git
   cd sms-spam-classifier

2. Install dependencies :

   pip install -r requirements.txt

3. Run the application locally :

   python app.py

   The application will be available at http://127.0.0.1:5000 if you have followed through correctly.

Usage

Local Testing

To classify a message, you can use the /api/label endpoint :

curl -X POST http://127.0.0.1:5000/api/label -H "Content-Type: application/json" -d '  {"message": "Congratulations! You have won a prize."}'

Preview

Here are screenshots of the application in action :

• Taking user input in the text box to classify an SMS message.

  

• Displaying the classification result as "HAM" or "SPAM" based on the input.

  

API Endpoint

POST /api/label

• Description: Classifies an SMS message as HAM or SPAM.

• Request Body: JSON object with a single key :

  {
    "message": "Your SMS message here."
  }

• Response: JSON object with the classification label :

  {
    "message": "Your SMS message here.",
    "label": "HAM" // or "SPAM"
  }

Error Handling

The application handles errors with custom templates :

• 404 Not Found : Returns a 404 page if the route does not exist.
• 500 Internal Server Error : Returns a 500 page for server-related errors.

License

This project is licensed under the MIT License.