Mining Analysis And Application Of E-commerce Data

With the rapid development of e-commerce,online shopping has become more and more frequent.Each e-commerce platform generates a large amount of transaction data and user comment data every day.These transaction data and user evaluation data are of great value to enterprises.Therefore,this paper mainly uses the crawling method to obtain the external data of the open enterprise from the e-commerce platform,and conducts research from the aspects of collection,cleaning,analysis and application,and uses the vacuum cleaner category of the Tmall platform as an example to do the analysis and application research.In this paper,the data collected by the Tmall platform vacuum cleaner is collected by the Locoy Spider.Due to the non-standardization of the collected Tmall data,it is necessary to clean and regularize the collected data.How to analyze and apply the data after the regularization has always been a big problem faced by the enterprise,which deserves our in-depth analysis and research.This paper mainly uses descriptive statistical analysis,comparative analysis,prophet time series prediction analysis,K-means cluster analysis and dictionary-based text analysis methods.In the industry market analysis and brand competition analysis,descriptive statistical analysis and comparative analysis are mainly used to obtain the market share,year-on-year growth rate,price and attribute performance of various vacuum cleaners in the industry,and also know the main sales categories of each brand and Price segment distribution.The Prophet model is applied to the trend prediction of vacuum cleaner sales,and the optimal parameters of each subcategory are found and predicted by the parameter optimization method.The prediction error rate of each sub-category is within 10%,and the prediction effect is better.It is predicted that the sales trend of wireless hand-held vacuum cleaners is better,while the horizontal vacuum cleaners have a downward trend.In the K-means brand cluster analysis,four categories were obtained,namely,a high-selling comprehensive brand,a high-priced sweeping robot brand,a high-priced wireless handheld vacuum cleaner brand,and a low-priced comprehensive brand.class.The purpose of brand clustering is to bring together similar brands,find real competitors to do competitive analysis,and identify strengths and weaknesses.In the text analysis of user comments,the comment analysis index system and the keyword corpus are mainly constructed in three aspects: buying experience,product experience and service experience.The user's concerns and emotional attitudes are analyzed by matching and analyzing user evaluations based on the keyword corpus.The product experience is most noticed by users,and the negative proportion of product experience is also the highest.The secondary indicators that users pay more attention to are noise,use experience,suction,quality,price and logistics.In addition,in the comparison between Midea and Haier,we find that Midea is better than Haier in noise and quality,and Haier is better than Midea in use experience,suction and price.Through the analysis of external industry data and competitors' data,we can understand the industry development trend,brand competition pattern and users' concerns and voicing points,better understand their own strengths and weaknesses,and know ourselves and others.Customer-centric improvements in products and services.