Network Intrusion Detection System. This is a classification problem where the model classifies the incoming connection request into two categories: normal (1) and anomaly (0). I used the KDD dataset available online which comprised 42 features and one target column. There was no missing value in the dataset, so I moved to the feature engineering part where I scaled the features and encoded categorical attributes into numerical. Then I did the feature selection by using a recursive feature elimination method. In the end, I partitioned the dataset and trained on four algorithms Decision Tree, KNN, logistic regression and Naive bayes. I got 99.6 accuracy on the decision tree. So, I saved that model and deployed it using flask.
ericblanco / fictional-happiness Goto Github PK
View Code? Open in Web Editor NEWLicense: Creative Commons Zero v1.0 Universal