基于MapReduce模型下FP_ growth算法的研究及应用

摘要

FP-Growth is a classic association rule mining algorithm which is based on frequent pattern tree. It has the highly application in many areas. Focus on the problem of the traditional FP-growth algorithm may produce lots of frequent item sets. This essay proposes a strategy of item sets mergering and pruning which can improve the process mining on FP tree. Thus, analyzing the mining method of single path and multi-path can reduce the number of parts of mining branch. Then the improved FP-Growth algorithm parallelized, based on the MapReduce programming technology. The experimental results show the improved method has an advantage in executing efficiency and it has better acceleration ratio and scalability.%FP_ growth算法是基于FP树挖掘频繁项目集的关联规则经典算法，在许多领域中有很高的应用价值。针对传统的FP_ growth算法可能产生大量的频繁项集，对FP树的挖掘过程进行了改进，提出了一种项合并剪枝的挖掘策略，进而分析了单路径和多路径的挖掘方法，减少了部分分支的挖掘次数。然后利用MapReduce模型，针对改进的算法并行化实现。实验结果表明该方法提高了算法的执行效率，并且具有良好的加速比和较好的扩展性。