农产品价格异常波动的市场价格传递与时空分布特征研究

The recent decade has witnessed a drastic fluctuation of agricultural product price. The frequent change of price has made significant impacts on the wellbeing of farmers, the sustainability of agriculture, and the food security of the entire nation. In response to these contingencies, mining mass increasing agricultural product price data is critical to monitoring the wellbeing of market and predicting imminent issues that might develop into large-scale food crisis. From the perspective of geography, the abnormality of agricultural product price and its influences on the market can be analyzed and presented by quantitative geographical methods, such as spatial-temporal autocorrelation, complex networks and 3D visualization. The main contents include： First, the project employs the technique of spatial autocorrelation to identify the pattern of agricultural product price at different geographical and temporal scales. A bivariate spatio-temporal autocorrelation model is established to account for both the market variables and price to study the spatial-temporal coupling rule. Second, the project identifies the spatial aggregation distribution and evolution rule of agricultural product price by implementing the spatio-temporal point pattern recognition method. Then the transmission model based on complex networks is established to summarize the transmission-diffusion characteristics among markets. Third, a 3D space-time model using kernel density function is built for improving the visualization of agricultural product price and highlighting the spatial locations and time periods in which the wholesale market of agricultural products may have involved abnormal events. This project applies cutting-edge spatio-temporal analyses to the case of abnormal fluctuation of agricultural product price in China. It provides a fruitful application of crowd sourcing big data while providing insight into practical challenges of abnormal fluctuation issues that China has experienced in recent years.

农产品价格的异常波动对农民收入、农业生产和国家粮食安全造成不利影响，如何从海量的农产品价格数据挖掘出有用信息是迫切需要解决的科学问题。本研究以近期价格波动幅度大频率高的农产品为例，运用时空数据挖掘、空间统计学、复杂网络、可视化等理论方法，定量分析农产品价格异常波动期间的时空关联性、聚集性和价格传递特征。主要内容包括：①研究不同时空尺度下农产品价格的空间自相关性、局部聚集特征，建立影响因素-农产品价格的时空自相关模型，归纳农产品价格与影响因素的时空耦合规律；②研究农产品价格时空点格局识别方法，分析价格的聚集特征和演变规律。建立基于复杂网络的农产品价格传递模型，挖掘价格在市场之间的时空传递特征；③建立农产品价格数据时空立方体模型，研究农产品价格时空分布三维可视化方法，实现数据的可视化表达。本项目一方面丰富和拓展时空数据挖掘的理论方法和应用领域，另一方面为农产品市场行情信息服务提供新的技术手段。

近年来，我国的猪肉、蔬菜等一些农产品呈现出价格波动频率增加、波动幅度加大的特点，价格的异常波动不仅造成农民生产经营的困扰，也给城乡居民生活带来冲击。本研究综合运用时空数据挖掘、空间统计分析、复杂网络和信息可视化等理论方法和技术手段，结合农产品价格的时空信息和产销属性等，以畜产品猪肉、蔬菜如黄瓜、西葫芦等为研究对象，研究了多尺度的农产品价格空间自相关性分析方法，构建了农产品价格与影响因素的双变量空间自相关模型，挖掘分析了农产品异常波动期间的时空分布格局、空间异质性和空间聚集性，从供给、需求和生产成本三方面测度与农产品价格的时空耦合关系，结果表明，城市行政区划级别越高，猪肉价格越低；在猪肉价格上涨较快时期，猪肉价格的空间自相关性随之减弱；导致猪肉价格剧烈波动的主要因素导集中在以生猪、仔猪和玉米为代表的成本因素方面；针对农产品价格横向传导过程复杂，现有方法难以揭示农产品价格波动传递动力学特征的问题，构建了基于时空影响域的农产品价格传递网络模型，利用复杂网络的特征参数对农产品价格传递过程进行分析，获得了农产品价格异常波动时期价格在各个市场间传递的网络结构、传递路径以及关键节点的空间分布与时序变化，为揭示农产品价格市场传递规律提供了一种有益方法；探究了基于点要素和面要素的农产品时空分布特征可视化分析方法和时空聚类可视化方法，为农产品市场行情信息的可视化分析提供技术手段。本研究深度挖掘和定量分析农产品价格异常波动时期的时空关联性、异质性、聚集性和市场传递特征，对于丰富和完善时空数据挖掘、复杂网络的理论和方法，拓展其应用领域具有重要的科学意义。