________________________________________ † : Mustafa Zaidi The 11th Asia Pacific Industrial Engineering and Management Systems Conference The 14th Asia Pacific Regional Meeting of International Foundation for Production Research Melaka, 7 – 10 December 2010 Laser Cutting problem modeling Using Statistical Tools Zaidi Mustafa † 1 and Bushra A. Saeed 2 Department of Computer Science, SZABIST Karachi, Pakistan Email: mustafainisb@gmail.com 1 bushra.saeed@szabist.edu.pk 2 Nukman Yusoff Department of Engineering Design & Manufacture, University of Malaya, Malaysia Email: nukman@um.edu.my I., Amin Department of Computer Science SZABIST Karachi, Pakistan Email: amin.imran@gmail.com Abstract - This paper describes the modeling of Laser cutting process of non-linear multivariable by using one and two way analysis of variance, linear and non linear regression analysis. The statistical techniques are used to explore better analysis techniques and improve the laser cutting quality by reducing process variations caused by controllable process parameters. The problem has already been solved by Taguchi-neural network method using one way analysis of variance. Orthogonal array used in Taguchi method is a very useful technique to reduce the time and cost of the experiment. The data set is very small in this method which causes difficulties in modeling and simulation of the problem. In classification problems decision tree is a very useful technique but it is not able to predict better results due to the small size of data. The results of analysis of variance are encouraging. Taguchi and regression normally optimizes input process parameters for single characteristics. The process industry most of the time needs to improve multiple quality parameters. One way and two way (interaction between two variables) analysis of variance can be used to predict better modeling compared to regression. Keywords: Laser cutting, Quality, Simulation of Laser cutting quality, Statistical Methods, ANOVA, Linear Regression 1. INTRODUCTION In this study, experimental analysis has been carried out to seek the optimum combination (laser power, cutting speed, assist gas pressure and standoff distance) of input parameters in laser cutting process in order to improve the laser cutting quality by other statistical methods on polystyrene foam to improve the deficiencies in Zaidi & Amin, (2010) statistical modeling. The observed values of edge quality, Kerf widths, percent overcut and material removal rate were measured for measuring quality. Only kerf width is used to understand the statistical techniques from (Zaidi & Amin, 2010). Orthogonal array were used to invest less time and cost in the experimentation. The effect of input parameters on output quality variation was assessed by different methods to determine the optimum input combination. Different techniques will be applied at this stage to understand these methods and compare them for better applications, viz, One way and two way ANOVA, single variable linear, multivariable linear, nonlinear and multivariable nonlinear regression Analysis. The software used was Excel and SPSS. Four controllable input parameters with their value divided into three stages to understand the effect of input