Open AccessKetidaktepatan Waktu Kelulusan Mahasiswa Universitas Terbuka dengan Metode Boosting Cart
Author(s) -
Gede Suwardika,
I Ketut Putu Suniantara,
Ni Putu Nanik Hendayanti
Publication year - 2019
Publication title -
jurnal varian
Language(s) - English
Resource type - Journals
ISSN - 2581-2017
DOI - 10.30812/varian.v2i2.361
Subject(s) - boosting (machine learning) , cart , decision tree , artificial intelligence , machine learning , classifier (uml) , voting , gradient boosting , decision tree learning , computer science , majority rule , random forest , pattern recognition (psychology) , engineering , politics , law , political science , mechanical engineering
The classification tree method or better known as Classification and Regression Tree (CART) has capabilities in various data conditions, but CART is less stable in changing learning data which will cause major changes in the results of the classification tree prediction. Predictive accuracy of an unstable classifier can be corrected by a combination method of many single classifiers where the prediction results of each classifier are combined into the final prediction through the majority voting process for classification or average voting for regression cases. Boosting ensemble method is one method that combines many classification trees to improve stability and determine classification predictions. This research purpose to improve the stability and predictive accuracy of CART with boosting. The case used in this study is the classification of inaccuracies in the Open University student graduation. The results of the analysis show that boosting is able to improve the accuracy of the classification of the inaccuracy of student graduation which reaches a classification prediction of 75.94% which previously reached 65.41% in the classification tree.