| |
Click here for full text:
Visual Mining of E-Customer Behavior Using Pixel Bar Charts
Hao, Ming C; Ladisch, Julian; Dayal, Umeshwar; Hsu, Meichun; Krug, Adrian
HPL-2001-144
Keyword(s): pixel bar chart; visualization; behavior
Abstract: To date, one of the common problems e-business and e- service want to solve is how to use their large volumes of sales histories, web transactions, and information resource data to know their customers' behavior. They want to understand their customers to develop long-term relationships and to improve service quality. Many data analysis methods use bar charts which only show highly aggregated data and lose most of the detail information due to data aggregation. This paper discusses the integration of a recent pixel bar chart technique that does not lose information into a visual data mining system. Our experiments show that the system allows the discovery of a wide range of patterns and trends. For example, it helps business analysts to identify the most profitable customers, the most frequent search key words, and the best sales timeline. Notes: To be presented at the ACM KDD-2001 Visual Data Mining Workshop, 26-29 August 2001, San Francisco, CA
7 Pages
Back to Index
|
printable version
