Algoritma Genetika: Studi Kasus Masalah Multi-Criteria Decision Analysis (MCDA) dalam Hal Ada Data Kosong

Abstract

Many methods on Multi-Criteria Decision Analysis (MCDA) are used to rank the alternatives based on the criteria . MCDA data can be presented in a decision matrix containing , the value in the -th alternative and -th criterion. The solution on MCDA methods is obtained by giving , weighted value on the -th criterion which is suitable with its role. The optimization concept of Spearman’s correlation in every pairs of solution candidate with all of criterias as a measure of goodness of the solution using genetic algorithm seems to be an alternative solution method for MCDA, even though, it is assumed that each criterion vector on matrix should be positively correlated. It is indicated from the results of the simulation against 30 alternatives with 15 criterias, genetic algorithm provides a solution that is high correlated with a result using the AHP method, the correlation of 0.94. Besides the treatment of missing value will be much simpler to use genetic algorithms and the result will be a high correlation between the ranking of alternative simulation from the complete data with alternative rankings contained missing value as much as 10% to 40%; all correlations were worth more than 0.85. A case study of 29 automobile brands with 11 criteria and contains 20% of missing value resulting Model-Y, Model-1, and Model-3 as the best sequence of three consumer preferred brands.