一种改进的模糊时间间隔序列模式算法

Sequential pattern mining is of great significance for discovering frequent patterns of transactions. With the requirements of data mining accuracy, sequential pattern mining has developed many research directions. Among them, sequential pattern mining with time intervals has a large number of application scenarios in real life, such as bank pipeline data analysis, Web log analysis and so on. However, at present, there are some problems in the research of mining sequential patterns with time intervals, such as poor mining accuracy when the time interval gap is too large, especially when the critical value has to rely on human estimation. Therefore, this paper proposes a mining algorithm for sequential patterns with fuzzy time intervals, TiCMiner. The main idea of this algorithm is to divide the time interval by means of fuzzy clustering, after that Sequential pattern mining is carried out for the sequence with time interval. To avoid increasing the processing time, this paper improve the connecting and pruning methods for sequential pattern mining at the same time. Finally, the algorithm is tested in the real bank pipeline data, which shows that the algorithm improves the processing accuracy of the time interval without increasing the running time.