Survey on FiDoop: Data Hierarchy Mining of Frequent Item sets Using MapReduce

Existing parallel mining algorithms lacks some of the features like parallelization of sequential code, balancing load over cluster of computers, and distribution of data over computers in a cluster. To overcome these problems we have introduced an algorithm called FiDoop. Which uses the MapReduce techenique for parallel frequent itemsets mining algorithm? FiDoop algorithm uses the ultrametric tree pattern for storage of data, it does not uses the FP(frequent pattern) trees like existing algorithm uses it. In FiDoop, mining task is completed with the help of MapReduce technique. This technique incorporates three MapReduce job to mine the large amount of data conventionally and economically. In first MapReduce job , all frequent itemsets are discovered. In second MapReduce job, it removes infrequent itemstes. Third MapReduce is most important, it construct small ultrametric trees which is helpful to mine the frequent data conventionally and economically. FiDoop algorithm is implemented in Hadoop cluster. For high dimensional data, FiDoop-HD is used, it is an improved version of FiDoop. FiDoop algorithm is an improved version of FIUT(frequent itemset ultrametric tree) algorithm.