/Comparing-ML-For-DDoS

A comparison on the metrics that different ML algorithms give trying to predict DDoS attack conditions

Primary LanguagePython

Comparing-ML-For-DDoS

The project is a comparison of 3 different ML classifiers in their prediction accuracy and speed of DDoS attack conditions.

The classifiers

  1. OneVsRestClassifier
  2. MultinomialNB
  3. BaggingClassifier

After using these simple algorithms, they are inturn compared with ensemble algorithms taken a combination of the above 3 mentioned algorithms

The ensemble classifiers

  1. Voting Classifier
  2. Weighted Classifier
  3. Stacking Classifier

3 permutations of the weights of the base classifiers are taken for 3 readings of the metrics of the ensemble classifiers

Metrics used for comparison

Accuracy(%) Mean Standard Deviation Precision Recall F-Measure True Postive True Negative False Positive

Datasets used

I did not get the permission to upload datasets, five datasets were used each containing network data and DDoS conditions for one kind of protocol The protocols are: ICMP, LAND , TCPSYN, TCPSYNACK , UDP