4th International Conference on Intelligent and Fuzzy Systems (INFUS), Bornova, Turkey, 19 - 21 July 2022, vol.504, pp.208-218
Finding periodicity of the patterns from super-market data has been found to be an important data mining problem which many researchers encounter often. Such patterns reflect the buying nature of the customers in the super-market. There may be yearly, half-yearly, quarterly, monthly, daily, hourly or any other type of periodicity. In such patterns, it has been observed that the pattern is not exactly periodic that is the length of the time interval of frequency of a frequent itemset is not always equal. Furthermore, the time gap between successive time intervals of frequency of the frequent itemset is also not equal. However, there is a sufficient overlapping in the time intervals. In this situation, if the time intervals of frequency can be retained as a compact form it becomes fuzzy time interval describing the period of frequency of the frequent itemset. We designate the corresponding pattern as fuzzy periodic pattern. We propose here a method of discovering such patterns from the super-market dataset. The effectiveness of the method can be verified with the help of experiment conducted with a synthetic data collected from FIMI web site.