Performance evaluation of industrial enterprises via fuzzy inference system approach: a case study
Makaleler > Tam Makale · 2015
Yazarlar: Gözde ULUTAGAY, Fatih Ecer, Efendi Nasibov
Soft Computing, SCI Expanded İndekslerine Giren Dergi, Uluslararası, Scopus, 2015, cilt 19, ss 449-458
DOI: 10.1007/s00500-014-1263-3
Özet
The aim of this study is not only to give self-contained and methodological steps of data mining with its areas of applications, but also to provide a compact source of reference for the researchers who want to use data mining and fuzzy inference in their area of work. We construct a fuzzy inference system to predict the profit of the major 500 industrial enterprises of Turkey. For this aim, we use most of the data mining tools. First, we use fuzzy (Formula presented.)-means clustering algorithm and obtain the linguistic terms of the variables. Having used decision tree technique, fuzzy rules are revealed. Eventually, we compare various defuzzification strategies to obtain crisp prediction values of our fuzzy inference system. We can conclude that the prediction results of the smallest of maxima defuzzification strategy-based fuzzy inference system has circa 40 % smaller sum square error than that of classical regression model.