Penggunaan Adaptive Neuro Fuzzy Inference System untuk Pembentukan Data Fuzzy dalam Association Rules Mining

Abstract

Forest fires are susceptible happened in Indonesia. Some of the factors that influence their occurrences are weather and climate. Both of them will cause the increase or decrease of the number of hotspots. Previous studies have succeed to provide the relation between the fuzzy data of climatology and number of hotspots using Association Rules Mining (ARM) with membership functions for the attributes are determined by the expert. However, the results have some rules with low percentage of support and confidence and few rules are not in accordance with the expert knowledge. It could be caused by the membership functions provided by the experts are not optimal. In this study, the method used is Adaptive Neuro Fuzzy Inference System to generate membership functions that are used in the formation of data fuzzy. Then, the data is processed with association rules mining that produce rules with better support and confidence percentage than the previous work. It is also able to manage to create new rules that comply with the expert knowledge