基于数据挖掘的农产品质量特性波动溯源方法

针对传统的农产品质量特性波动溯源方法一直存在溯源不准确、效率低的问题,提出基于数据挖掘的农产品质量特性波动溯源方法。依据农产品质量特性之间的关系,将质量特性间的波动传播关系划分成链接内部传播、链接串联传播、链接并联传播以及链接混联传播四种基本形态,求出每种形态的波动传播度。通过小波分解方法对农产品质量特性波动特征进行提取,结合波动传播度为溯源提供可靠的技术支持。介绍了决策树方法,通过基于决策树方法将农产品质量特性波动传播度和Shannon熵看作特征,依据训练集形成规则并产生决策树,输入测试集进行数据挖掘,得到农产品质量特性波动溯源结果。实验结果表明,所提方法不仅具有良好的特征提取效果和较高的溯源精度,而且时间复杂性指标较低。

Agricultural products quality fluctuation traceability method based on data mining

For traditional agricultural products quality fluctuation characteristics of tracing method has been traceability is not accurate, the low efficiency problem is proposed based on data mining of agricultural products quality fluctuation characteristics of tracing method, according to the relation between the characteristic of agricultural product quality, quality characteristics of wave propagation in the division into internal communication link, link series communication, link parallel communication and link mixed contact spread four basic forms, for each form of the wave propagation, through the wavelet decomposition method for agricultural product quality fluctuation characteristics of feature extraction, combined with wave propagation degree to provide reliable technical support for traceability. The decision tree method is introduced. Through based on decision tree method will agricultural product quality characteristics of wave propagation index and Shannon entropy are treated as characteristic, according to the training set of rules and produce decision trees, input test sets for data mining, agricultural product quality fluctuation characteristics of the traceability. The experimental results show that the proposed method not only has good effect of feature extraction and high accuracy analysis, and low time complexity.