ANALYSING AND FINDING FREQUENT PATTERNS USING MULTIPLE MINIMUM SUPPORT THRESHOLD
JOURNAL OF MECHANICS OF CONTINUA AND MATHEMATICAL SCIENCES
Data mining is the process of discovering interesting patterns from the transactional database. In the past decade, numerous techniques have been proposed for mining frequent patterns using single minimum support threshold for all items from the transactional database which results in "rare item issue". While fixing the minimum support to higher level, it results frequent patterns where rare item are missed. While fixing the minimum support to lower level, it results in too many frequent
... any frequent patterns which is known as combinatorial explosion. To confront the rare item problem, an effort has been made in the literature to find frequent patterns with "multiple minimum supports thresholds". In this approach, minimum item support (MIS) is given to each item for mining frequent patterns. In this article, comparative analysis is done between MISFP-Growth and MISLP-Growth algorithm for mining frequent patterns using multiple minimum support threshold. In MISLP-Growth algorithm array based structure is adopted which is the major advantage and in MISFP-Growth algorithm pointer based structure is adopted which is the disadvantage. For this, the experiments are conducted using benchmark databases to find the efficient algorithm.From the results produced by these algorithms, it is found that the MISLP-Growth algorithm outperforms MISFP-Growth algorithm for all the databases in the criteria of consumption of runtime and memory.