Abstract:

The data sets used in scientific studies pose a very complex structure from time to time. At this point, data mining is making a big contribution in terms of improving the quality of services by revealing useful information from large databases. Generally on studies, to predict future data trends utilization of the methods, data mining techniques in one of the most widely used are classification and regression models. In this study, among data mining methods, classification and regression models most commonly used ones are decision tree algorithms. By comparing Classification and Regression Trees (CART) algorithm which belongs to decision trees and logistic regression shows classification characteristics on real data set and success rates of these two methods. In this context, taken by the Social Security Administration pharmacy provision system, from the respiratory disease which is one of 11 diagnoses for 6,772,313 entries in the prescribed antibiotics in the penicillin group was used to analyze that profiling the patients and the analysis found the CART analysis has better classification success than logistic regression analysis

Keywords: