Research On Big Data Anomaly Judgment Of Supermarket Chains Based On FCE

With the rapid development of the global information society,the retail industry is producing data all the time,and the need to deal with and analyze the massive data has also increased sharply.The retail business competition is fierce,for the managers of retail enterprises,it is necessary to quickly and accurately find out the abnormal business situation from the data and make the business adjustment quickly so that the enterprise can continue to develop.The evaluation of retail enterprises by customers on the Internet,that is,network public opinion,is also of great referential value to enterprises.Therefore,the enterprise needs to synthesize the financial data in the field and the network public opinion data in the off-the-counter to judge whether the management data is abnormal or not,and analyze the causes of the anomalies to adjust the strategy so as to improve the core competitiveness of the enterprise.Abnormal detection of business conditions is a challenge in the operation of retail enterprises.In order to meet this challenge,two problems should be solved first.The first problem is how to integrate network public opinion data with financial data.The second problem is that the conventional machine learning anomaly detection algorithm applied to multidimensional mixed management data detection of chain supermarket is not effective,what method should be used to apply to complex data of chain supermarket,so it is urgent to solve these two problems.According to the characteristics of supermarket data,this paper analyzes the feasibility of the existing anomaly detection algorithm,and optimizes and improves the traditional fuzzy comprehensive evaluation method(FCE).A dynamic FCE model is proposed to judge the abnormal operation data of supermarket chain and run it on the distributed computing framework of Spark.The main contents of this paper are as follows:(1)Pre-processed financial data in chain hypermarket and cleaning and removing dirty data.Obtain and quantify the over-the-counter public opinion data of supermarket chain,and establish the data set after integrating with financial data.(2)Aimed at the disadvantage of traditional FCE artificial weight value error,this paper establishes the FCE anomaly judgment model of mixed network public opinion chain supermarket(NPO-FCE),and optimizes the original FCE subjective weighting method by triangular fuzzy number-analytic hierarchy process(TFN-AHP).Improve the accuracy of abnormal judgment results of supermarket management data.(3)Aimed at the disadvantage that the FCE can not adapt to the market change dynamically,this paper establishes a dynamic supermarket FCE anomaly judgment model of mixed network public opinion,which combines the correlation between factors and the deviation degree of related factors.The factor weight set and the factor membership set of FCE are generated adaptively and dynamically,the human error is removed,the rate of misjudgment is reduced,and the accuracy and credibility of abnormal judgment of supermarket management data are improved.This paper makes use of the real management data set of BBK commercial chain Limited by Share Ltd.The experimental results show that the two abnormal judgment models can be used to analyze the abnormal judgment of chain supermarket management data.