/DDoSAttacks

Primary LanguageJupyter Notebook

GDP

In this project, using several machine learning algorithms, we are attempting to find the best model that can detect any DDOS attack that a server experiences.

At this point of time, we are done with data cleaning, feature engineering and training models that we are planning to choose the best out of. Before cleaning the data we had 88 features which after feature engineering we scaled down to the 15 most significant features. All the models were trained using all the features pre- feature engineering and their performance metrics were calculated. The process was repeated for only those features that we got after feature selection. Results vary before and after feature engineering for each models. Some showed a drastic change whereas some performed very similar. However, Random Forest and Decision Tree surpassed other models that were trained in both cases. We are planning to train the models further by adding and removing featuers based on their score making sure no significant features are left out and no unnecessary features are beign used.