| Summary: | 碩士 === 國立中正大學 === 資訊管理學系暨研究所 === 101 === Sequential pattern mining (SPM) is one of most important data mining technique, and it is widely used in customer behavior scenario. Organizations are able to explore customers’ purchase habit and comprehend the relationship between merchandise and each other through SPM process to develop sales policy.
However, all customers’ sequence is regarded as with the same importance in the past research. But in reality, each customer generally has a different significance to organization in most of case. Therefore, weighted sequential pattern mining (WSPM) has been coming up to discuss. Customer lifetime value (CLV) is a remarkable way to distinguish between customers in terms of their importance. Nevertheless, there is no research about using CLV to do WSPM until now. This research utilize different CLV model to calculate weights of sequences and adjust Prefixspan algorithm to do WSPM. In experiment, we test runtime, number of patterns, value per pattern, precision, recall, F-measure to compare performance between our method and traditional SPM.
|
|---|
