Author : Nikhil Prakash Gorde and Prof. B.B.Gite

To discover the frequent itemset is very significant task in data mining. These frequent itemsets are beneficial in applications like Association rule mining as well as co-relations. To mine frequent itemsets these systems are with certain algorithms, but these are incompetent in allocating as well as balancing the load, when it comes through extreme data. Automatic parallelization is also impossible with these algorithms. To overcome these issues of present algorithms there is necessity to develop algorithm which will support the missing features, such as spontaneously parallelization, balancing as well as good distribution of data. In this paper we are with a new method to discover frequent itemsets by using MapReduce. Modified Apriori algorithm is used with HDFS environment this is called FiDoop Method. In this technique mapreduce process will work individually as well as simultaneously by using the decompose strategy. The outcome of this mapreduce method will be given to the reducers then reducers will display the outcome. In the experiment we used three diverse algorithm like basic apriori, FP Growth and our proposed modifies apriori, the system has executed in standalone machine as well as distributed environment and shown the results how proposed algorithm is better than existing algorithms.

Keywords: Association Rules, Frequent item sets, FiDoop, MapReduce, Modified Apriori.