A Spam Classifier for SMS/Emails

Databases : SMS Spam Collection Dataset (https://www.kaggle.com/datasets/uciml/sms-spam-collection-dataset)

Technologies/Frameworks: Python, Pandas, Numpy, Sklearn, streamlit (Frontend Framework), matplotlib, nltk, seaborn,

run spamClassifier.ipynb and get pkl files and add to your project workplace

install required packages & run streamlit run main.py

Steps for this project:

1. Data Cleaning
   • renaming
   • missing values
   • remove duplicates
2. EDA (to understand underlying data)
   • plotting charts (ham vs spam)
   • wordcount
3. Text Pre-Processing (with the help of nltk library)
   • Lower case
   • Tokenization
   • Removing special characters
   • Removing stop words and punctuation
   • Stemming
4. Model Building (with the help of sklearn library)
   • train-test data
   • tfidf vectorization
   • model training on various ML classifiers
5. Evaluation
   • compare and choose on best model
6. Improvement
   • re-train model by hyper parameter tuning (here TfidfVectorizer(max_features=3000))
7. Website
   • create & open a project in editor
   • crete & code app.py file
   • import .pkl files & functions from ipynb file
   • integrate it in streamlit

Local URL: http://localhost:8501

Network URL: http://192.168.1.113:8501

Example Outputs: