An Algorithm for Generating Non-Redundant Sequential Rules for Medical Time Series Data

Abstract

In this paper, an algorithm for generating non-redundant sequential rules for the medical time series data is designed. This study is the continuation of my previous study titled �An Algorithm for Mining Closed Weighted Sequential Patterns with Flexing Time Interval for Medical Time Series Data� [25]. In my previous work, the sequence weight for each sequence was calculated based on the time interval between the itemsets.Subsequently, the candidate sequences were generated with flexible time intervals initially. The next step was, computation of frequent sequential patterns with the aid of proposed support measure. Next the frequent sequential patterns were subjected to closure checking process which leads to filter the closed sequential patterns with flexible time intervals. Finally, the methodology produced with necessary sequential patterns was proved. This methodology constructed closed sequential patterns which was 23.2% lesser than the sequential patterns. In this study, the sequential rules are generated based on the calculation of confidence value of the rule from the closed sequential pattern. Once the closed sequential rules are generated which are subjected to non-redundant checking process, that leads to produce the final set of non-redundant weighted closed sequential rules with flexible time intervals. This study produces non-redundant sequential rules which is 172.37% lesser than sequential rules.