Research On Soybean Quality Comprehensive Evaluation Based On Combination Weight And K-means Clustering

With the sustained and stable development of the economy and society,the types and quantities of agricultural products in China have continued to increase,and the agricultural product market has gradually improved.The consumer demand for agricultural products in China has gradually increased from a "quantitative safety type" to a "good quality type".People pay more attention to the quality of agricultural products.The evaluation and improvement of the quality of agricultural products has aroused the attention of scientific research workers and consumers.The evaluation and improvement of the quality of these products can enrich china’s agricultural product quality evaluation system and provide references for agricultural workers to select and breed agricultural products.Aiming at the above-mentioned problem of agricultural product quality assessment,this thesis proposes a soybean quality evaluation model according to the combined weighting and K-means clustering algorithm,explores and researches its establishment and application.By analyzing the current situation of agricultural product evaluation,this paper summarizes three aspects of agricultural product evaluation,including evaluation methods,the determination of attribute weights,and the follow-up processing of evaluation results.According to the research steps of multi-attribute evaluation,firstly,the factors affecting the quality of soybean are analyzed,and various indicators in the evaluation system were determined.Then,by comparing the advantages and disadvantages of subjective weighting and objective weighting methods,it is determined to select the combination weighting method combining the order relationship analysis method and the entropy weight-CRITIC method to obtain the weight of each index in the soybean evaluation system.This not only effectively avoids the uncertainty of the subjective judgment of experts in the evaluation process,but also combines the inherent nature of the sample data,and the weight setting is more reasonable.Finally,the scoring results of eight soybean samples were obtained and analyzed in detail.In terms of result processing,the unsupervised learning k-means clustering algorithm was used to analyze the soybean samples that participated in the evaluation,and it was found that the clustering results were consistent with the previous scoring results regardless of whether they were divided into two or three categories.This verifies the rationality of the evaluation results and provides a scientific and effective method model for soybean quality evaluation.