Detecting DDoS Attacks Using ML and Random Forest Feature Classifier Method

Computing in the cloud allows customers to access services over a network whenever they need them. Use our services anytime and wherever you want. Security issues plague the paradigm, which has various uses. Cloud availability of services and safety is affected by DDoS assaults. DDoS detection is necessary for real users’ access. Several studies have increased dataset accuracy to differing degrees. This article presents a cloud-based DDoS detection method. This research focuses on reducing misclassification error to improve DDoS detection accuracy. The suggested work selects important features using the mutual information (MI) and random forest feature importance (RFFI) techniques. The characteristics that were chosen are then processed using gradient boosting (GB), random forest (RF), weighted voting ensemble (WVE), logistic regression (LR) and K-nearest neighbour (KNN). Experimental findings indicate RF, GB, WVE as well as KNN accuracy of 98% with 20 characteristics. Analysing misclassifications of these approaches leads to more accurate measurements and further investigation. Extensive testing show the RF effectively detects DDoS attacks, misclassifying just single attack as common. Results are compared to verify the suggested approach.