Improving Tree-Based Machine Learning (ML) Classifier Techniques for Web Mining: An Empirical Evaluation

In order to better recognize and categorize consumer behaviour in internet-based applications, this look at gives an intensive analysis of using tree-based machine learning category strategies in web mining. Web mining is an essential subset of records mining that examines user browsing styles and conduct through taking useful facts out of net server logs. The paper examines and assesses the performance of three classifiers which are tree based type algorithms: TreeJ48, Random Tree, and REP Tree, within the classification of web utilization patterns. Our method contains an intensive experiment conducted on one-of-a-kind dataset percentage splits (60, 70%, 80%, and 90%) to evaluate each algorithm's accuracy and error quotes. The outcomes display that, for all splits, the TreeJ48 method plays better than the others in phrases of classification accuracy and the lowest rates of misclassification.