📱 SMS Spam Classifier

An end-to-end NLP pipeline that automatically flags spam in SMS messages using classical machine-learning models.

---

🔍 Overview

1. Data

   • Source: SMS Spam Collection v1
   • Format: label\tmessage (ham/spam)

2. Preprocessing

   • Strip URLs, emails, numbers, punctuation
   • Expand contractions, lowercase
   • Tokenize, remove stop-words, lemmatize

3. Feature Extraction

   • TF-IDF vectors (unigrams & bigrams)

4. Modeling & Tuning

   • Train/test split (80/20)
   • Compare 7 classifiers:
     • Multinomial Naïve Bayes
     • Logistic Regression
     • Decision Tree
     • Random Forest
     • Gradient Boosting
     • Support Vector Machine
     • k-Nearest Neighbors
   • Select best by accuracy

5. Evaluation

   • Confusion matrix
   • Accuracy summary

6. Usage

   import joblib
   model = joblib.load("models/best_model.pkl")
   print(model.predict([
       "Free entry in 2 a wkly comp to win FA Cup finals!", 
       "Are we still meeting at 6pm?"
   ]))
   # ['spam', 'ham']