OLAP-based Scalable Profiling of Customer Behavior
Profiling customers’ behavior has become increasingly important for many applications such as fraud detection, targeted marketing and promotion. Customer behavior profiles are created from very large collections of transaction data. This has motivated us to develop a data-warehouse and OLAP based, scalable and flexible profiling engine. We define profiles by probability distributions, and compute them using OLAP operations on multidimensional and multilevel data cubes. Our experience has revealed the simplicity and power of OLAP-based solutions to scalable profiling and pattern analysis.
KeywordsCustomer Behavior Fraud Detection Calling Pattern Incoming Call Dimension Duration
Unable to display preview. Download preview PDF.
- 1.S. Agarwal, R. Agrawal, P. Deshpande, A. Gupta, J.F. Naughton, R. Ramakrishnan, S. Sarawagi: On the Computation of Multidimensional Aggregates, Proc. VLDB’96 (1996)Google Scholar
- 2.Q. Chen, U. Dayal, Meichun Hsu: A Distributed OLAP Infrastructure for E-Commerce, International Conference on Cooperative Information Systems (CoopIs99), UK (1999)Google Scholar
- 3.H. Garcia-Molina, W. Labio, J. Yang: Expiring Data in a Warehouse, Proc.VLDB’98 (1998)Google Scholar
- 4.S. Geffner, A. El Abbadi, D. Agrawal, T. Smith: Relative Prefix Sums: An Efficient Approach for Querying Dynamic OLAP Data Cubes, Proc. ICDE’99 (1999)Google Scholar
- 5.J. Han: OLAP Mining: An Integration of OLAP with Data Mining, Proc. IFIP Conference on Data Semantics (DS-7), Switzerland (1997)Google Scholar
- 6.Q. Chen, U. Dayal, Meichun Hsu: Multidimensional and Multilevel Calling Pattern Similarity Comparison, Tech Rep. HP Labs, 1999.Google Scholar