Mining Negative Associations from Frequent and Regular Patterns through Application of Maximal Property

Knowledge hidden in databases normally is mined to find frequently occurring Item-Sets and then move on to find the positive associations that exists among those patterns. It is challenging to fix frequency threshold for finding such patterns. The spread of the item sets could be huge even though the Frequency of occurrence of the Item-sets is high, Regular occurrence of the Item sets is also very important while achieving high threshold on the frequency. Considering Both frequency and regularity may involve in generation of more item sets than expected. Most of the focus is on finding the positive associations that exists among the frequently and regularly occurring patterns. Sometime negative associations that exeunt among the frequently and regularly occurring patterns have high significance in the medical field. Reducing Item-sets to a minimum number will be helpful in finding the most accurate and important associations from Frequent and regular frequent Item-set. Maximal property can be applied such that minimum patterns that occur frequently and regularly that represents whole lot of patterns can be found. These pruned patterns can be used for finding effective negative associations. This paper presents an algorithm that finds negative associations that exists among the frequent and regular patterns through application of Maximal property. The Algorithm is implemented on a sample e-commerce data and by using IBM supplied 100K data and the accuracy of finding the negative associations that exists among the Frequent and regular Item-sets is nearing 99%