Implementation of classification and regression Tree (CART) and fuzzy logic algorithm for intrusion detection system

Intrusion detection system is a system that can detect the presence intrusion or attack in a computer network. There are 2 type of intrusion detection system that misuse/signature detection and anomaly detection. This research use a combination of Classification and regression Tree (CART) and Fuzzy Logic method is used to detect intrusion or attack. CART is used to build rule or model that will be implemented by fuzzy inference engine. Testing process is performed using Fuzzy Logic without doing defuzzification because the resulting rule will be used as a classification. Training, testing and validation of the model is done by using KDD Cup 1999 dataset that has been through the preprocessing and cleaning data process. Accuracy testing and validation is calculated by using the confusion matrix. From several test performed, the best model is built from training 70%, the depth of tree 11 and node leaf minimum percentage 90% with an accuracy was 85,68% and average time validation was 21,92 second.