https://github.com/inddrsingh/email-sms-spam-classifier

Given a text, the ML model can predict whether it is "SPAM" or "NOT SPAM"
https://github.com/inddrsingh/email-sms-spam-classifier

machine-learning-algorithms naive-bayes-classifier python3 tfidf-vectorizer vectorization

Last synced: 4 months ago
JSON representation

Given a text, the ML model can predict whether it is "SPAM" or "NOT SPAM"

• Host: GitHub
• URL: https://github.com/inddrsingh/email-sms-spam-classifier
• Owner: INDDRSINGH
• Created: 2025-03-16T14:04:07.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2025-09-07T16:05:19.000Z (9 months ago)
• Last Synced: 2025-10-03T22:25:55.323Z (8 months ago)
• Topics: machine-learning-algorithms, naive-bayes-classifier, python3, tfidf-vectorizer, vectorization
• Language: Jupyter Notebook
• Homepage: https://email-sms-spam-classifier-meag.onrender.com/
• Size: 1.09 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# Email-SMS-Spam-Classifier

![DatabaseSchema](https://github.com/INDDRSINGH/Email-SMS-Spam-Classifier/blob/main/email%20logo.jpg)

##  Overview  

This project uses machine learning to classify given text as **“SPAM”** or **“NOT SPAM”**. Built using Python's NLP tools and deployed via a web interface, it demonstrates a complete pipeline—from data ingestion to model deployment.

##  Project Goals  

- Train a reliable spam classifier to enhance communication filtering  

- Showcase core analytics and ML skills: data preprocessing, model evaluation, and deployment  

- Deliver an interactive application accessible through a web browser

##  Tools & Technologies  

- **Python**  

- **Jupyter Notebook** for experimentation  

- **Pandas & Numpy** for data handling  

- **NLTK** (for stemming/tokenization)  

- **TF-IDF Vectorizer** for feature extraction  

- **Multinomial Naive Bayes** classifier  

- **scikit-learn** for model training and evaluation  

- **Streamlit** for web app (via `app.py`)  

- **Render** for cloud deployment  

- **Pickle** for model persistence (`model.pkl`, `tfidf.pkl`)

##  Project Flow  

1. **Data Loading & Exploration**  

   - Loaded dataset from Kaggle, performed exploratory analysis (text length, word frequency, class balance)  

2. **Preprocessing**  

   - Cleaned text (lowercasing, punctuation removal)  

   - Applied stemming using NLTK  

   - Vectorized text using TF-IDF  

3. **Model Training**  

   - Evaluated Multiple ML algorithms  

   - Selected Multinomial Naive Bayes as the best-performing model  

4. **Evaluation**  

   - Metrics: accuracy, precision, recall, F1-score (found in Jupyter Notebook)  

   - Visualized results with confusion matrix  

5. **Deployment**  

   - Built a Streamlit app to expose prediction interface  

   - Deployed live to Render (viewable at the project’s GitHub link)

##  Live Demo  

Try the classifier live [here](https://email-sms-spam-classifier-meag.onrender.com)

## Dataset

- **Source**: Kaggle ([Link](https://www.kaggle.com/datasets/naserabdullahalam/phishing-email-dataset?select=CEAS_08.csv))

- **Contains labeled text messages/emails (spam vs. ham)**

## Key Insights

- Stemming helps reduce vocabulary size and improves model generalization

- TF-IDF provides a strong representation behind Naive Bayes for text classification

- The classifier shows high precision on spam detection (specific metrics available in notebook)

## What I Learned

- Hands-on experience with NLP preprocessing and feature engineering

- Practical model selection and evaluation methods

- Building and deploying a production-ready ML web app using Streamlit

- Managing end-to-end ML workflow from Jupyter to live deployment