Voting Classification Method for Network Traffic Prediction

Abstract

Prediction analysis (PA) is a data mining-based technique. The futuristic outcomes based on present data can be predicted using this technique. As the dataset is large and complex, network traffic classification is a big concern in prediction analysis. Three phases are included in network traffic strategies. The data collection is obtained in the first step of pre-processing, and it is analysed to delete incomplete and obsolete values. The relationship between function and goal set is formed in the second process. In the final step, the classification technique is used to classify the data. The various intrusion attacks on the internet, and also the methods to detect them, also motivated this research work. We reviewed and analysed the well-known network traffic data,NSL KDD dataset and its various features in this research. The proposed model combines Logistic Regression and K-nearest neighbour classifiers with a voting classifier to differentiate data into malicious and non-malicious classes with higher efficiency than current models.