<![CDATA[The aim of the proposed research is to compare and test the accuracy of data mining classification algorithms. Comparing algorithms that depend on different parameters of a given data set. There are learning and classification algorithms that are used to analyze, study and classify the available data. However, the problem is finding the best algorithm and the desired results with the highest level of accuracy in predicting future values or events from a data set. Where the classification models used are the C4.5 and Naïve Bayes algorithms. Testing and validation using k-fold Cross Validation as well as evaluating the performance of the prediction model using the ROC-AUC graph with graphic visualization. The data used as samples were taken from ATM machine problem data with a total of approximately 250 samples. Testing was carried out with the help of the Rapidminer tool with operators and parameters used in creating models of the algorithms being compared. The tests that have been carried out prove that the C4.5 algorithm has the best performance with an average accuracy value of 96.00%, a recall value of 97.78% and a precision value of 92.14%, while the naïve Bayes algorithm produces an accuracy value of 83. 00%, the recall value is 76.40% and the precision value is 84.82%. Apart from that, evaluation and validation in this test is also seen based on the ROC curve called AUC (Area Under the ROC Curve) where for the C4.5 algorithm the value is 0.931 while naïve Bayes is 0.894 so the C4.5 algorithm is categorized as Very Good Classification because it has a value between 0.90-1.00. These results show that the C4.5 algorithm is proven to be a potentially effective and efficient classification algorithm.]]>
