T-product Quality Evaluation Based On Data Mining

In the process of product quality evaluation,sampling inspections,questionnaire surveys and other methods are commonly used.Although these methods can reflect the quality status to a certain extent,there are still small sample sizes,high sampling costs,difficult to find unknown threats,and non-conformity with consumer evaluations.And other issues.At this stage,with the development of the Internet and data mining technologies,it is convenient and feasible to obtain real evaluations of consumers on products.According to the backgrounds above,this article evaluate the product quality of T products(children's wooden toys)from comments with data mining technology.During the research process,the article first proposed an understanding of the concept of product quality.Secondly,this paper adopts the Word2Vec and KNN method in data mining to propose the T product quality index extraction algorithm.Based on this algorithm,T products the quality evaluation index is constructed.Then,the article takes this evaluation index as the characteristic,training product quality evaluation model based on supports vector machine.For the imbalance problem of data sample,the article introduces cost sensitive factor to modify the initial model.The revised model has a better overall prediction effect and the prediction ability of negative samples is improved.Based on this evaluation model,the article evaluates the overall product quality and quality indexes of T products by using the 183,875 comment data on Tmall and Jingdong platform.In the end,the paper makes a comparative analysis of our evaluation and the results of quality sampling,and puts forward relevant suggestions based on the analysis from the aspect of quality supervision work and the e-commerce platform management.The main conclusions of this paper are as follows:(1)The accuracy of T product quality index extraction algorithm is 76.23%and 71.64%,which is relevantly effective to extract the product quality index.(2)The overall prediction effect of the SVM quality evaluation model based on the cost sensitive is better,and the F1 value of the predicted evaluation index of the rare negative sample is increased from 63%to 72%.(3)The quality of product quality of T products is 80.72%,which is lower than the quality of sampling inspection.(4)According to the attention and quality,this paper divides each quality index into four categories:high attention and high quality,high attention and low quality,low attention and low quality,low attention and high quality.(5)Based on the results of data analysis,it provides suggestions for quality supervision agencies and e-commerce platforms.The article applies the data mining method to the product quality evaluation,furflier enriches tiie research on data mining and product quality.In practice,the T-product quality evaluation model which can be applied to evaluate product quality has certain practical value Based on the results of comparative analysis,suggestions are put forward for quality supercision agencies and e-commerce platform to assist the current product quality supervision and management.