COMPARISON OF TREE BASED CLASSIFICATIONS AND NEURAL NETWORK BASED CLASSIFICATION

In this paper, we empirically analyze and compare the performance of neural network-based classification (MLP) and decision tree based classifications (CART and Random Tree) on data sets with banking and medical purpose information. We included structural parameters for distinguishing the classification methods. We also introduced up sampling and down sampling along with feature selection and found a more detailed analysis of the Precision, Recall, F1 score, Area under the Curve, Test and Train accuracies which are necessary in order to judge the performance of the learning and classification method of all the models. However, we identified some limitations involved with these sampling techniques during the research such as loss of vital data and overfitting outcomes.