Download PDF

How low can you go? The value of sparse data in retail databases

Paper
First Online:
Revised:

Abstract

In this paper, the authors examine the effect of sparse data in retail databases on the ability of logistic regression, CHAID and neural networks to produce predictive models. The density of a large retail sales data set was systematically increased through the elimination of low-density cases and the resulting impact on prediction examined. The results suggest that market analysts should consider dropping low density cases when dealing with retail sales data and other data sets composed partly of sparse data.

Keywords

Database Marketing Customer Relationship Analytical CRM e-CRM Direct Mail Telemarketing Targeting Segmentation Behavioural Analysis Systems Profiling Campaign Intelligent e-marketing Interactive Market Modelling eCommerce Internet Information Management Research Data Protection 
Download to read the full article text

Copyright information

© Palgrave Macmillan 2002

Authors and Affiliations

  1. 1.F.C. Manning Chair in Economics and Business at the Dalhousie School of Business Administration
  2. 2.Completing an interdisciplinary PhD at Dalhousie University researching Marketing Informatics

Personalised recommendations