计算机技术与发展2017，Vol.27Issue(3)：108-112,5.DOI:10.3969/j.issn.1673-629X.2017.03.022
基于云计算的SPRINT算法研究
Research on SPRINT Algorithm Based on Cloud Computing
摘要
Abstract
Decision tree is a very important technology in data mining,which is often used for data analysis and forecasting. When the tra-ditional decision tree algorithm is dealing with massive data mining,the CPU and memory is limited,resulting in its shortcomings like long time-consuming,poor fault tolerance and small storage capacity. Faced with massive data processing,cloud computing has a lot of advantages in this respect. It places emphasis on the good algorithm of SPRINT. First of all,it is optimized,and then parallelized in order to make the optimized algorithm better applied to cloud computing. When traditional SPRINT algorithm generates the decision tree,multi-valued bias problem will happen,and when it generates a node,through the calculation of Gini index of two layer,the effects of multi-valued bias is reduced. In parallel algorithm,through the distribution of data to the processor execution,then collecting and processing,the total time of execution is reduced. The experimental results show that the improved SPRINT algorithm based on cloud computing platform has better classification accuracy,and at the same time,its execution speed gets obvious improvement.关键词
云计算/MapReduce/SPRINT算法/Gini指数Key words
cloud computing/MapReduce/SPRINT algorithm/Gini index分类
信息技术与安全科学引用本文复制引用
杨洁,黄刚..基于云计算的SPRINT算法研究[J].计算机技术与发展,2017,27(3):108-112,5.基金项目
国家自然科学基金资助项目(GZ211018) （GZ211018）
