Use of classification and regression tree (cart) for chronic periodontitis classification in dental hospital of Hang Tuah University Surabaya

Abstract

Classification and Regression Tree (CART) is now commonly used as one of nonparametric
classification methods. The advantage of this method is it can effectively used on data which
have many independent variables. It is also easy to be interoperated.
The aim of this research is to classify Chronic Perodontitis in Dental Hospital of Hang Tuah
University Surabaya using CART. The Chronic Periodontitis divided into 3 categories: mild,
moderate, and heavy. Based on the analysis result, 2 factors can affected the mild type of
chronic periodontitis , 6 factors affected moderate type of chronic periodontitis and 8 factors
affected the heavy type of chronic periodontitis.
The criteria used to count the accuracy of the CART classification is 1-APER. The 1-APER
score shows a very good result which is 92,5% for data learning and 90% for data testing.