MANAGING PRODUCTION PROCESS IN A PET RESIN INDUSTRY USING DATA MINING AND GENETIC PROGRAMMING

Abstract

Balancing the production volume and costs because of the petrol prices and, thus, supply change rapidly is one of the managerial issues in the polymer industry. In this study, data about the chemical operation of PET resin products have been used, and monthly and annual production plans and their influencing factors have been analyzed with data mining techniques in the plastic industry. The algorithm of find-dependencies used to find effective parameters has been identified in the levels of monthly production, sale, and end-of-the-month inventory. Rules have been established with the find-laws algorithm, and genetic programming is used to predict the outputs. It shows that high-accuracy applicable rules can be obtained with these technics. The rules proved to be more accurate at the end of the comparison and became employable for decision-support to the production process of the PET Resin factory. With these techniques used for the first time on such a problem in the literature, all similar companies can provide clarity in their strategic decisions and efficiency in using company resources and production.