Using parallel approach in pre-processing to improve frequent pattern growth algorithm

Mining frequent itemset is an important step in association rule mining process. In this paper we are applying a parallel approach in the pre-processing step itself to make the dataset favorable for mining frequent itemsets and hence improve the speed and computation power. Due to data explosion, it is necessary to develop a system that can handle scalable data. Many efficient sequential and parallel algorithms were proposed in the recent years. We first explore some major algorithms proposed for mining frequent itemsets. Sorting the dataset in the pre-processing step parallely and pruning the infrequent itemsets improves the efficiency of our algorithm. Due to the drastic improvement in computer architectures and computer performance over the years, high performance computing is gaining importance and we are using one such technique in our implementation: CUDA.