Research On The New Automobile Retail Model Based On Data Mining

With the sharp decline sales of automobile production and the severe competitiveness among the industries,each automobile enterprise began to pursue the user satisfaction and profit of post-market.Due to the increasing pursuit of personalization by users and the increasing demand for products,the automotive aftermarket is facing higher challenges.A new retail model of automotive aftermarket that takes user demand as the core and highly integrates online and offline has injected new vitality into the auto industry.The new retail mode of online purchase and offline experience not only greatly facilitates the purchase process of users,but also breaks the regional and time restrictions of traditional 4S stores,making consumer prices more transparent and technical services more standardized.Generally,users like to post online comments after offline experience.By analyzing these review data which contain a lot of emotional attitudes,users’ concerns can be deeply mined and problems of products can be found,so as to monitor brand reputation and product quality.This paper selects the user comment data of "oil type" automobile service in Jingdong Mall,and fully combines the text mining technology and machine learning related theories and methods to carry out emotional classification and text clustering analysis on the text data.First,12000 pieces of user comment information of "oil type" automobile service were collected through Python crawler technology,and the data obtained are preprocessed,mainly including text de-duplication,Analytic compression to word,Chinese word segmentation,word stopping and so on.Then,TF-IDF algorithm was used for word frequency statistics and word cloud analysis.Next,semantic analysis and emotional analysis were performed on the processed text data through ROSTCM6 and Python Snow NLP library to distinguish positive tendency and negative tendency,Machine learning classification algorithm was used to build the auto service comment classifier.Finally,Kmeans algorithm and LDA theme model are used for text clustering analysis,and LDA theme models are constructed for positive and negative texts respectively.By analyzing high-frequency keywords and potential themes in the three themes,advantages,disadvantages and improvement methods of products and services are found,so as to improve the competitiveness of enterprises.