Title: Development of customer usage profile histograms

Authors: Sheng-Kuei Lin, Lynn E. Hall, Jeffrey H. Stilman

Addresses: Customer Correlation and Test Schedule Development, DaimlerChrysler Corporation, 800 Chrysler Drive, CIMS 481-01-27, Auburn Hills, MI 48326, USA. Customer Correlation and Test Schedule Development, DaimlerChrysler Corporation, 800 Chrysler Drive, CIMS 481-01-27, Auburn Hills, MI 48326, USA. Customer Correlation and Test Schedule Development, DaimlerChrysler Corporation, 800 Chrysler Drive, CIMS 481-01-27, Auburn Hills, MI 48326, USA

Abstract: Customer use profiles provide crucial information for product design. A good understanding of customer behaviour can help engineers improve their designs. The key problem for customer profiling is that we want to identify that level of customer that will lead to a successful and cost effective design. The upper (lower) limit does not identify the specific customers we are looking for. It just provides the largest (smallest) value of the collected data. In this paper, a parametric (Weibull) technique and a non-parametric (series representation through Fourier series) technique are applied to identify the customer group of interest. An example is presented for each method. The proposed non-parametric approach successfully identifies more characteristics of customer behaviour than the Weibull approach.

Keywords: 95th percentile; customer profiles; Fourier series; nonparametric; parametric; Weibull distribution.

DOI: 10.1504/IJMPT.2001.001267

International Journal of Materials and Product Technology, 2001 Vol.16 No.4/5, pp.393-403

Published online: 01 Jul 2003 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article