• 

  Data Visualization

About the Project

Inspiration

The increasing volume of spam messages inspired me to develop an AI-powered SMS Spam Detection system that accurately classifies ham (legitimate) and spam (unwanted) messages.

What I Did

I implemented Natural Language Processing (NLP) and Machine Learning (ML) techniques to build an efficient spam classifier. The key steps included:

• Data Preprocessing: Used NLTK, Regex, and feature engineering to clean and prepare the dataset.
  
• Exploratory Data Analysis (EDA) & Visualization: Analyzed data distribution and patterns.
  
• Feature Extraction: Applied Bag of Words (BoW) and TF-IDF for text representation.
  

• Model Training & Evaluation:

  • Gaussian Naïve Bayes with BoW achieved 88% accuracy.
    
  • Gaussian Naïve Bayes with TF-IDF also achieved 88% accuracy.
    
  • Random Forest with BoW achieved 97% accuracy.
    
  • Random Forest with TF-IDF delivered the highest 98% accuracy.
    

Challenges Faced

1. Selecting the best feature extraction technique for optimal classification.
   
2. Handling imbalanced data to improve model performance.
   
3. Fine-tuning ML models for better spam detection accuracy.
   

Conclusion

This project successfully demonstrates the effectiveness of ML and NLP in spam detection. Random Forest with TF-IDF provided the best results with 98% accuracy, while Random Forest with BoW achieved 97%, and Gaussian Naïve Bayes performed consistently with 88% accuracy for both BoW and TF-IDF. This highlights the impact of feature extraction and model selection in text classification.

Built With

• jupyternotebook
• machine-learning
• matplotlib
• natural-language-processing
• nltk
• numpy
• pandas
• python
• regex
• scikit-learn
• seaborn