Using Data Mining and Three Decision Tree Algorithms to Optimize the Repair and Maintenance Process

Abstract

The purpose of this research is to predict the failure of devices using a data mining tool. For this purpose, at the outset, an appropriate database consists of 392 records of ongoing failures in a pharmaceutical company in 1394, in the next step, by analyzing 9 characteristics and type of failure as a database class, analyzes have been used. In this regard, three decision tree algorithms have been used to determine the most important attributes and to determine the effective rules for the failure. Based on the results obtained from the feature selection of all the three used algorithms, the lifetime characteristics of the machine, the name of the machine and the duration until the last failure are recognized as the most important attributes. On this basis, the life of the device has a very special importance. Since the depreciation in the pharmaceutical industry is high, so the life of the devices used in maintenance and repair has a special effect. In this regard, machines with a life span of more than 20 years are subject to high depreciation and failure, and in addition to the usual repairs they need some special repairs.