This Machine Learning Spam Classifier uses to LogisticRegression to classify messages/mails. Even though i didn't do parameter tuning or try to balance out the difference in the number of spam and ham messages we have with imblearn.over_sampling(SMOTE), the model ended up with a test accuracy of 98%. How does it work? - This classifier takes messages from the historical spam and ham messages, evaluates it over and over again till it picks the difference in what a spam or ham message is.