Classification Model On Household Clothing Expenses In The City Of Ambon Using Multivariate Adaptive Regression Splines (MARS) Methods

Abstract

Traditional market as one of the shopping facility has a special place in our society. Nowadays, the people have
a change of lifestyle: In the past they only had a traditional market, but now they also have a modern market. This
improvement changes the way people shopping. The purpose of this study is to identify a relationship between how
people choose traditional or modern market to buy their household clothing and the variable that affect those decision
using classification method. A good classification method should give the least misclassification rates. Multivariate
Adaptive Regression Spline (MARS) model is one kind of classification method which often used when there are many
categorical variable of response, the data doesn’t have a pre specification model, and the predictor variable consist of
categorical and continuous data type. This research purpose is to get the best model on classification, taking the case
study of household preference on clothing shopping place in Ambon, and its influencing variables, based on 2012’s Cost
of Living Surveys (SBH). Method performance is measured by its accuracy rate, Noise Signal Ratio (NSR) and G-Mean
from classification table.
Keywords: Household clothing expenses, MARS, Cost Of Living Survey