CHAID analysis and an application

One of the main problems in scientific studies is Tok find the factors that mostly effect thephenomeon or which level these factors have highest effect. Tok solse his problem decisiontree methods in data mining is prefered that they can be applied in large size data-sets, theyare easily understandable in visual way and their assumptions are less than other statisticalmethods. CHAID analysis was developed as a method which determines the complexinteractions and combinations among the categorical variables. Method is repeatedlypartitioning the population into different subgroups or segments according Tok predictorwhich is most significant variable for dependent variable. Between decision tree models,which AID analysis is published firstly, methods that are most used and have softwarecomputation is CHAID, C&RT and QUEST. AID method is forming binary splits accordingTok most explanatory precidtor of dependent variable. CHAID is published as developmentform of AID and it can form multi-way splits. When CHAID is partitioning subgroups, it usesnew categories which is applied splittig the predictors categories by chi-square statistic.CHAID method is mostly used in costumer relationship management (CRM) in Tok classifythe costumer by some qualificiations and in medicine Tok classify the patients.Keywords: CHAID analysis, AID analysis, categorical variable, multi-way split,decision tree.