SALES FORECASTING WITH MODEL TREES METHOD: AN APPLICATION IN STAINLESS STEEL SECTOR
MODEL AĞAÇ YÖNTEMİYLE SATIŞ TAHMİNİ: PASLANMAZ ÇELİK SEKTÖRÜNDE BİR UYGULAMA

Author : Orhan ECEMİŞ
Number of pages : 336-350

Abstract

Decision trees are used in many areas including classification, regression problems. The main difference of the Model Tree method from other regression trees; instead of producing an estimated value in the leaves, it creates multivariate linear models. In this study, the sales analysis and estimation of a company operating in stainless steel sector were realized by using M5P (Model Tree) method. Total sales, montly sale values of machinery and metal sectors between January 2008 and December 2017 were obtained from the database of the company. The independent variables in the model were determined as nickel, chrome, iron, prices, dollar rate, Producer Price Index, Industrial Production Index and Manufacturing Industry Capacity Utilization Rate, which affect stainless steel demand. According to the findings obtained, four models were obtained according to the Dollar Dry-Chromium mine critical values, which are similar to each other in terms of total sales and metal sector sales. Root node and critical values in the Total Sales and Metal Sector sales were realized in different figures. Six models were obtained according to the PPI-Dollar dry critical values, which are root knot PPI in the estimation of the machinery sector.

Keywords

Decision Trees, Sales Forecasting

Read: 637

Download: 220