Email Spam Detector Project: Project Overview

This project uses Natural Language Processing and Classification to predict whether an email is a spam or not.

• Create a tool that classify email spams to help users identify spam emails
• Engineered features from texts to bag-of-words model and tf-idf, in order
• Used Naive Bayes Classifier to reach a high accuracy

Code and Resources Used

• Python Version: 3.7
• Packages: pandas, numpy, matplotlib, seaborn, scikit learn
• Kaggle Data: https://www.kaggle.com/datatattle/email-classification-nlp

About The Data

The data contains the following columns:

• Message Numbers
• Message Bodies
• Labels

EDA

I looked at the distributions of the data and the value counts for the various categorical variables. Below is the highlight from the pivot tables.

Model Building

I split the data into train and test sets with a test size of 20%. I pre-processed the texts using Count Vectorizer. I then created a function to removed stops words and punctuations from the texts. I turned the texts into bag-of-words model and transformed them into tf-idf, in order. I used Naive Bayes Classifier and evaluated it using Accuracy. I chose Accuracy because it is relatively easy to interpret.

Model Performance

• Naive Bayes Classifier: Accuracy = 0.92