基于K-Means聚类的农产品价格异常数据检测①讲述.PDFVIP

2017 年 第 26 卷 第 3 期 计 算 机 系 统 应 用 基于 K-Means 聚类的农产品价格异常数据检测① 韩 琳, 吴华瑞, 顾静秋 (北京农业信息技术研究中心, 北京 100097) ( 国家农业信息化工程技术研究中心, 北京 100097) (农业部农业信息技术重点实验室, 北京 100097) (北京市农业物联网工程技术研究中心, 北京 100097) 摘 要: 全国各地各个年份的农产品市场价格数据量庞大, 而海量的农产品的市场价格数据中无可避免存在超 出市场正常价格范围的异常价格元素, 这对搜索引擎农产品市场价格的统计分析与预测造成了影响. 从市场价 格大数据中发现离群点并计算出价格边界成为有待解决的问题, 为此, 本研究在数据挖掘聚类技术K-means 算法 的基础上, 提出了基于 K-means 聚类的农产品市场价格异常数据检测并计算出农产品市场价格边界, 测试及实践 结果表明该方法提高了聚类的精确率和稳定性, 实现了价格异常点检测与价格边界的计算. 关键词: 海量农业数据; 聚类; K-means 算法; 离群点; 市场价格; 异常检测 Abnormal Agricultural Price Data Detection Based on K-Means Clustering HAN Lin, WU Hua-Rui, GU Jing-Qiu (Beijing Agricultural Information Technology Research Center, Beijing 100097, China) (National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China) (Key Laboratory of Agricultural Information Technology of Ministry of Agriculture, Beijing 100097, China) (Research Center of Beijing Agricultural IOT Engineering Technology, Beijing 100097, China) Abstract: Vertical search engine of the ministry of agriculture needs to collect the market price data of agricultural products in various years from all over the country. It can not be avoided that the massive agricultural market price data has abnormal price point, which has an impact on the analysis and forecast of the agricultural market price. It needs to be solved to find market price data outliers and calculates the price boundary. Therefore, on the basis of the traditional data mining clustering K-means algorithm, this study achieves the outlier data detection and calculation of the boundary of the price of agricultural products, test and practice results show that the method improves the clustering accuracy and stabil