摘要
针对供应链金融信用评估中传统单机环境和简单机器学习预测分析的高效性和精准性问题,研究了一种并行的集成学习评估模型,阐述了模型基于MapReduce并行的编程模型,集成多个随机森林,通过建模和预测实现信用评估的理论和技术实现原理。分析了模型的供应链金融信用评估性能,得出该模型能为大数据环境下的供应链金融信用评估提供可行参考的结论。
As to the predicting efficiency and accuracy in traditional single-machine environment and simple machinelearning methods while applied to supply chain financial credit assessment,a parallel ensemble learning assessmentmodel is studied.This paper elaborates the principle of MapReduce parallel programming method which integratesmultiple random forests to implement credit assessment by modeling and predicting.And it analyzes the performance ofproposed model while applied to supply chain financial credit assessment,and comes to the conclusion that proposedmodel could provide a viable reference method for achieving parallel credit assessment in the circumstances of big data.
作者
杨灵运
邓生雄
严芸
Yang LingYun;Deng Shengxiong;Yan Yun(Guizhou CASICloud-tech Co., Ltd., Guiyang Guizhou 55000)
出处
《现代工业经济和信息化》
2017年第17期50-53,共4页
Modern Industrial Economy and Informationization
基金
2016年贵阳国家高新区科技创新创业局"云供应链大数据金融服务平台软件"资助项目
作者简介
杨灵运(1984—),男,硕士,工程师,研究方向:工业互联网、智能制造、云制造、区块链技术、工业大数据等;邓生雄(1988—),男,硕士,助理工程师,研究方向:工业大数据、机器学习、工业互联网等;严芸(1989—),女,硕士,工程师,研究方向:工业大数据、机器学习、工业互联网等