1. INTRODUCTION The main objective of this study was to investigate applicability and constrains of the linear (MLRA) and nonlinear (ANN) methods of statistical analysis. Both approaches were used for modeling the same data set obtained during experiment facilitated under the industrial conditions. The technological process that was the target of this investigation was aluminate solution decomposition, as the part of the Bayer alumina production process. However, technological process in question is not that much of interest in this paper. This paper is focused on the methodological process of modelling. Same methodological approach could be used for many other industrial processes. In the process which was modeled in these investigations, following input variables were considered: concentration of the Na 2 O (caustic); caustic ratio and crystallization ratio; starting temperature; final temperature; average diameter of crystallization seed and duration of decomposition process. Only one output variable was correlated above defined inputs COMPARISON OF LINEAR AND NONLINEAR STATISTICS METHODS APPLIED IN INDUSTRIAL PROCESS MODELING PROCEDURE Predrag Đorđević, Ivan Mihajlović* and Živan Živković University of Belgrade, Technical faculty in Bor Vojske Jugoslavije 12, 19210 Bor, Serbia (Received 10 May 2010; accepted 10 June 2010) Abstract This paper presents the comparison of Multiple Linear Regression Analysis (MLRA) and Artificial Neural Networks (ANN) as the statistical analysis tools. Most influential statistical parameters for choosing right modeling tool are evaluated in this investigation. Investigation was performed on real statistical data set obtained after measurements of the process parameters underindustrial conditions. Keywords:MLRA, ANN, statistical modeling * Corresponding author: imihajlovic@tf.bor.ac.rs Serbian Journal of Management Serbian Journal of Management 5 (2) (2010) 189 - 198 www.sjm06.com