Have you ever wanted to learn how to detect scams in emails? In this Github, I go in-depth on building different machine learning models to do just that. This Github requires a basic knowledge of Python and Pandas to be able to build the model. We will use Sci-kit Learn's machine learning library to build algorithms to detect whether the email is a scam or not. We will be using UCI's Scambase Dataset, which includes a list of features and a classification if it is a scam (1) or not(0). You can see the data and the data definitions through the Scambase website.
The results of the models are below:
As you can see, the Second Random Forest Classification, XGBoost, and AdaBoost Classifier were very close and had an accuracy of 95%. The #1 predictor I found for classifying emails as spam or not spam is "char_freq_%21".
Medium Article: https://medium.com/analytics-vidhya/building-machine-learning-models-to-detect-scams-in-email-41ef3bd86bdb