| With the development of science and technology,as the main means of transportation for people to go out,the energy and power system of automobiles have been constantly changing.Nowadays,the pollution of the ecological environment and the excessive consumption of natural resources have attracted more and more attention,and the development of the new energy vehicle industry is also expanding.Today,with the prevalence of e-commerce,various auto portals have emerged.As a very common channel for people to express their opinions on the internet,online reviews can reflect the true feelings of consumers about new energy vehicle products.Through text mining and emotional analysis on a large number of comment texts,it is of practical significance to know users’ feedback of new energy vehicles while they use the car,and offer proposals for companies to optimize their products.This thesis firstly uses the method of web crawler to collect the online comment data of new energy vehicle users in Pacific Auto Network,and perform data preprocessing on it.Through the exploratory analysis of the data,the user profiles of new energy vehicles are given from the city,model and price distribution of car buyers.According to the parameter configuration information,245 new energy vehicle models are clustered and analyzed,and finally pure electric new energy vehicles and plug-in hybrid electric new energy vehicles are grouped into four categories respectively,which provides some references for potential consumers when purchasing vehicles.Secondly,this thesis uses the TF-IDF algorithm to extract the feature words in the review text,and preliminarily selects the nouns related to the attributes of new energy vehicles as candidate words for satisfaction factors.Based on the Word2 Vec model,the word vector of each feature word is calculated,and the K-means clustering method is used to cluster them into 6 categories,which are summarized as:exterior design,interior decoration,driving performance,handling experience,interior space and supporting devices,as the second-level indicators for evaluating user satisfaction,and finally a total of 28 feature words were selected as the third-level indicators for satisfaction evaluation.According to the TF-IDF value of each three-level index,the weight of each index is determined,and a new energy vehicle user satisfaction evaluation system is constructed.Logistic regression,Support vector machine and LSTM models are established to classify the sentiment of the review data.Finally,the Support vector machine model is selected to calculate the praise rate of each three-level indicator with better effect,and then obtain the user’s overall satisfaction score for new energy vehicles.This thesis uses the quadrifid graphs model,according to the satisfaction and importance,the various influencing factors of the overall satisfaction of new energy vehicle users are analyzed,and some suggestions are given.Finally,five popular new energy vehicles were selected: Hongguang MINIEV,Model 3,Xiaopeng P7,Tang New Energy and Qin New Energy,and calculated their user satisfaction scores,at last the scores were compared.The results show that the user’s overall satisfaction score for new energy vehicles is0.7332(total score 1),which is at an upper-middle level.Users gave a large degree of affirmation to its external design and control experience,but the satisfaction score in the dimensions of interior space and supporting devices was the lowest,mainly due to the dissatisfaction with the storage function of the trunk and the sound insulation of the car.high.Through comparative analysis,it is found that the Hongguang MINIEV has the highest user satisfaction,and its advantages are reflected in the interior and handling,and the battery life needs to be improved.The satisfaction score of the Tesla Model 3 is at the bottom,mainly due to certain differences in workmanship,space and sound insulation. |
