Research On Methods Of Anomaly Detection Of Commodities' Logistic Master Data Under E-Commerce

For e-commerce companies,improving the efficiency of order fulfillment and reducing operating costs are the main means to improve their competitiveness.For this reason,e-commerce companies are paying more and more attention to the use of information technology in their logistics services.Information technology requires high-quality data as a foundation.In logistics services,we are most concerned about the categories,sizes,volumes,and weights of commodities,which we call the commodities' logistic master data.If these data are wrong,it will affect the subsequent warehousing and transportation operation process and operation efficiency.It may cause economic losses and economic disputes for enterprises,carriers or transporting and distribution personnel.Commodities' logistic master data is usually collected manually when the commodities are in the warehouse,there will be a certain percentage of errors.Today's e-commerce companies sell a large number of products,up to millions.In order to reduce erroneous data,it is necessary to efficiently find the erroneous data as accurately as possible.This article researches the anomaly detection method of commodities' logistic master data.The thesis firstly analyzes the characteristics of the commodities' logistic master data by using exploratory data analysis methods based on the data of company G.Then we searched for related papers on anomaly detection,and based on the characteristics of e-commerce commodity logistics master data,we proposed a data anomaly detection method to deal with the erroneous commodities' logistic master data.The method includes three steps:dimension sorting,extreme abnormal data detection and suspicious data detection.The boxplot method and the isolation forest are used together to finally realize the anomaly detection of commodities' logistic master data.At last,we use G company's commodities' logistic master data information to verify the effectiveness of the anomaly detection method proposed in this article,and introduce the deployment and application of this method in G company.