Penerapan Algoritma Tree Augmented Naive Bayesian pada Penentuan Peubah Penting

Abstract

In the era of free market competition today, improving product quality is very important. Consumer
preferences through product level of analysis is one method that many manufacturers conducted to
evaluate the product. Multivariable regression is a statistical method used to determine the important
variables. The weakness of this method is the strict assumption. This problem will be completed by
the method of bayesian networks. There are several algorithms to build the BN. This study uses TAN
and NB because of its simplicity. This study shows that the most accurate method at the chosen level
of classification accuracy is the TAN by 83%. The importance variable is the aspect liking of strength
of after taste.