Using SWAT(Secure water treatment testbed) data to predict when the system is under attack. We are using 6 known types of attacks to apply machine learning algorithm . We are applying feature selection method to reduce the number of features from 52 to 10 using Chi square method. Selecting 10 features from PCA as we found out that keeping 10 top features will retain 99.97% of data information. Then we apply ensemble method to train and test data using different ML algorithms on test data. The data we observed are:
Test Accuracy: 0.9993762993762993 Val Accuracy: 0.9993761696818465 Train Accuracy: 0.9993761696818465
Test Accuracy: 1.0 Val Accuracy: 1.0 Train Accuracy: 1.0
Test Accuracy: 1.0 Val Accuracy: 1.0 Train Accuracy: 1.0
Test Accuracy: 0.9991683991683992 Val Accuracy: 0.9989602828030776 Train Accuracy: 0.9990642545227698
Test Accuracy: 0.9985446985446985 Val Accuracy: 0.9983364524849241 Train Accuracy: 0.9984404242046163
Test Accuracy: 1.0 Val Accuracy: 1.0 Train Accuracy: 1.0
Test Accuracy: 0.9993762993762993 Val Accuracy: 0.999168226242462 Train Accuracy: 1.0
Test Accuracy: 0.7796257796257796 Val Accuracy: 0.7797878976918278 Train Accuracy: 0.7797878976918278
Test Accuracy: 0.8667359667359668 Val Accuracy: 0.866916198793928 Train Accuracy: 0.8668122270742358
Test Accuracy: 1.0 Val Accuracy: 1.0 Train Accuracy: 1.0