Research And Application Of Automobile User Portrait Based On Improved K-means Clustering Algorithm

In the era of big data,data mining technologies widely used in various fields are constantly changing.More and more enterprises are committed to subdivide users and dig deep into their needs,and strive to maximize the commercial value of data.From the wave of car making in the past few years,it is not difficult to see that the new forces of car making in the head have generally adopted the strategy of building popular models and seizing the market,which has a strong impact on the marketing environment of traditional car enterprises.The reason for the strong development momentum of emerging auto enterprises is that they have taken advantage of the favorable policies of the state to promote the intelligent networking of automobiles and the cleaning of automobile energy;But more importantly,the effective use of existing and potential user data by the new power of car making is also the key to its surprise.Aiming at different travel needs and fully meeting the target audience,the welldesigned popular products can be described as sharp and precise,attracting a considerable number of user groups.It can be seen that the driving habits of the target user group are effectively and accurately distinguished,so as to tap the potential needs of automobile users,and then provide them with appropriate services from an appropriate perspective.These problems deeply related to the user experience are interrelated,which makes the driving habits of automobile users become a key research topic in the industry,and the mining and analysis of automobile user information is particularly important.However,the information security awareness of car users and the private space attribute of cars make it difficult to select and collect sample features.Therefore,the core problems that need to be faced in the research process are as follows: firstly,according to the characteristics of automobile users,the sample data and its collection method that can reflect the driving attitude of automobile users are designed;Then select the appropriate optimization ideas and improve the algorithm suitable for the clustering research of automobile user samples,so as to obtain the reliable driving posture portrait of automobile users and relevant conclusions.In order to solve the above problems,the following methods are proposed in this study:（1）This paper uses P2 SOM neural network with improved weight adjustment algorithm.After fully exploring the basic principle,network structure and the weight adjustment rules of neuron nodes in the competitive layer of SOM（self-organizing feature map）neural network,aiming at the problem that the gap between automobile user data samples is relatively small and easy to cause the pendulum effect,the weight adjustment principle of forgetting the second place is proposed,and an innovative algorithm is designed,It is used to select the K value and initial clustering center required in the K-means clustering algorithm.（2）This paper implements the principle of maximum distance to select the cluster center.Considering that car users may have some extreme preference when adjusting seats and steering wheel,which will lead to outlier sample points.Therefore,after carefully studying the small to medium to large distance algorithm,this paper draws lessons from its idea of widening the distance between cluster centers as far as possible,and always uses the maximum distance principle as the selection strategy of cluster centers from the selection of the first cluster center to the end of the algorithm,so as to avoid the situation that the K-mean clustering algorithm falls into the local optimal solution due to outliers as far as possible.The comparative experiments of the above two algorithms are carried out with public data sets.After analyzing the experimental results,it is decided to use the former with better clustering effect of the two methods to study the driving habits of car users.Therefore,this paper takes the collected automobile user data as the sample,uses the improved P2SOM-K-means algorithm to carry out the clustering experiment,and analyzes and verifies the results in SPSS software.The results show that the clustering effect of each type of data is remarkable,and the samples in the data set are evenly divided into five clusters.It is concluded that P2SOM-K-means algorithm is competent for the clustering research of automobile users’ driving habits.According to the clustering results,the car users are divided into five categories according to their driving habits,and then the car users are clustered and evaluated.According to the car user groups with different types of driving habits,this paper provides corresponding driving posture adjustment suggestions and safety precautions.