Performance analysis for skewed data

Information technology and the media have changed the face of business practices today. Customers now play the key role for the success of any business and noncompliance to their specifications will simply lead to failure of the business. Researchers across many disciplines have worked out several modifications of the traditional process capability measures to obtain better estimates of the products capability to meet market specifications. However, these conventional capability measures heavily depend upon the theory of normality. In this paper, we compare and contrast the Cumulative Distribution Function (CDF) method with the latest proposed process capability evaluation methods such as Burr percentile method and commonly used Clements percentile method when the underlying distribution is non-normal. A simulation study using Gamma, Weibull and Beta distributions is conducted and the comparison of the results is presented. Finally, a case study is presented using actual data from a manufacturing process.