摘要
农产品价格数据量日益增长,从中挖掘隐含有用的信息十分重要。以2010年至2012年波动较大的猪肉价格为例,采用空间统计方法挖掘分析全国23个省市价格的时空分布格局,研究农产品价格波动时期的空间自相关特征,研究不同的空间权重矩阵和空间统计量对空间自相关性的影响。结果表明,在猪肉价格稳定期间,各批发市场的价格总体呈现空间集聚格局;而在价格超过20元/kg时期,各批发市场的价格没有明显的相关性;从局域上看,价格稳定期间沿海发达地区高值显著聚集,价格高峰时期有64%的市场价格位于"高-低"或"低-高"区域;不同空间权重矩阵下,Moran’s I和Geary’s C反映的空间自相关性结果一致。空间统计分析能够挖掘价格数据的空间特征,并通过GIS的可视化手段使挖掘结果更加直观,是农产品价格数据挖掘的有效方法。
At present the agficultural market information increases dramatically. It＇s important and significant to dig out hidden useful information from the mass data. The objective of this study was to explore the application of spatial statistical methods in analysis of prices of agricultural products. Taking pork prices from January 2010 to August 2012 for example, the spatial statistics methods were adopted to analyze the spatial distribution pattern of prices in 23 provinces. The spatial dependence of price during violent fluctuation period and effects of weight selection on spatial autocorrelation statistics were discussed. Global spatial autocorrelation analysis results reveal that prices have an overall clustering pattern during the stable period, but when the price is more than 20 yuan/kg, the correlation of price is not obvious. Local spatial autocorrelation analysis shows that the high price gathered in coastal areas when the price is stable. The 64% of markets belong to ＂high - low＂ or ＂low - high＂ area during the peak period. By comparing different weights matrix, Moran＇s I and Geary＇s C reflect the consistent results of spatial autocorrelation. The spatial characteristics of price data is mining by spatial statistical analysis, and mining results are more intuitive through GIS visualization means, so the spatial statistical methods is an effective method of mining agricultural market information.
出处
《农业现代化研究》
CSCD
北大核心
2014年第1期29-32,共4页
Research of Agricultural Modernization
基金
国家科技支撑计划"农村农业信息资源整合关键技术集成与应用"(编号:2011BAD21B02)
北京市自然科学基金"农田渐变环境自组织网络覆盖控制方法研究"(编号:4122034)
关键词
农产品价格
数据挖掘
空间统计
空间自相关
agricultural products＇ price
data mining
spatial statistics
spatial autocorrelation
