随机决策树改进算法在大数据上的设计与实现

With the rapid development of Internet technology, huge amount of data has been accumulated. Using machine learning algorithm to discover the underlying value of data has become a research hotspot. However, due to the growth of the data size and dimension, machine learning algorithm are facing unprecedented challenges in terms of efficiency and calculation cost. Random decision tree algorithm can reduce the calculation cost by building trees without using any purity detection. Because of the tree structure, it makes the algorithm too difficult to parallel deal with the large data problems. In this paper, we propose an improved random decision tree algorithm, using LSH algorithm instead of the tree structure to randomly shatter the data. The improved algorithm is non-iterative, so it can flexibly and efficiently parallel compute in the distributed platform with high accuracy compared with the original algorithm. As a result, the improved algorithm can meet the storage and processing requirements of big data.