: Data mining is a new methodology for extracting information in scientific decision-making process and often uses statistical and machine learning techniques for data analysis. A new approach by combining statistical methods and machine learning for analysis of huge data gains more information than using them separately. In this thesis, data mining process, logistic regression and decision trees method are introduced and with combining CART, an algorithm of decision trees and multinomial logistic regression, a new method is presented for analysis of categorical data. This technique is applied to real data for steel production line and the results show that the analysis of data is more effective and informative particularly on ordinal responses.