Customer Activity Sequence Classification for Debt Prevention in Social Security

摘要: From a data mining perspective, sequence classification is to build classifier using frequent sequential patterns. However, for complete set of patterns on large dataset can be extremely time-consuming and the number discovered also makes pattern selection building very time-consuming. The fact that, in classification, it much more important discover discriminative than set. In this paper, we propose novel hierarchical algorithm classifiers Firstly, mine which are most strongly correlated each target class. step, an aggressive strategy employed select small Secondly, pruning serial coverage test done mined that pass used sub-classifier at first level final classifier. And thirdly, training samples cannot covered fed back stage with updated parameters. This process continues until predefined interestingness measure thresholds reached, or all covered. generated loop form Within framework, searching space reduced dramatically while good performance achieved. proposed tested real-world business application debt prevention social security area. shows effectiveness efficiency predicting occurrences based customer activity data.