P-IOC-FPMine: an association rule mining algorithm for item constraints based on the MapReduce framework

Abstract

In the application of big data, one of the most challenging problems is how to consider the requirements of users. To avoid this problem, we proposed IOC-FP-growth. This added user-defined pre-term or post-item constraints into the classic FP-growth algorithm. What's more, a parallel data mining algorithm based on MapReduce, namely P-IOC-FPMine was proposed, which was a low-memory fast association rule mining algorithm. Finally, by evaluating the effectiveness of the method in the public data. The results showed that the IOC-FP-growth method can consider the user's needs for association rules easily. Compared with the FP-growth, Recorder and PNARCMC, we only need to meet the requirements of users to extract accurate rules, which will bring huge advantages for data mining. After parallelisation, the performance of the P-IOC-FPMine was better than FP-growth on the data set. The results showed that the P-IOC-FPMine was more appropriate for handling large-scale data sets.