COMPARING LINEAR REGRESSION AND ARTIFICIAL NEURAL NETWORKS TO FORECAST TOTAL PRODUCTIVITY GROWTH IN IRAN

In the recent years we have seen widespread discussion about productivity. The most common measures of productivity that are widely used by economists and business analysts, are production productivity, labor productivity, industrial workshop productivity and total productivity. The concept of total productivity is defined as real output per unit of all inputs. Presumably, total productivity growth reflects phenomena such as general knowledge, the advantages of particular organizational structures or management techniques, reductions in inefficiency, and real locations of resources to more productive uses. Given the importance of this concept, in this paper we use a multiple regression equation and an Artificial Neural Network (ANN) model to forecast the total productivity based on three indexes such as labor productivity, production productivity and industrial workshops productivity from 2006 until the 2012 M.D. Then, we compare these two methods according to five criteria and finally we conclude that the artificial neural network model is better than linear regression method to forecast the total productivity but the difference is not very significant. Data that used in this study are several time series which represent production productivity, labor productivity, industrial workshop productivity and total productivity from 1979 until the 2005 M.D. They are published in Statistic Center of Iran.