Research on New Energy Vehicle Customer Intelligent Mining Model Based on Principal Component Analysis and Goal Planning Algorithm

Since the data set contains multiple distributions of sub-data sets. Firstly, anomaly detection based on density is used to process the outliers. Secondly, the Collinearity Diagnostics of all variables finds that the independent variables have multicollinearity,therefore, we consider using principal component analysis to eliminate the influence of multicollinearity of independent variables, and using Spearman correlation coefficient to remove some variables. Then, we obtain important influencing factors of brand purchasing power through principal component analysis. Thirdly, unpurchased customers are digging deeply, and the customers who are most likely to buy are selected from each brand as the target customers, and the service difficulty coefficient p and the service difficulty degree are defined to describe the role of service difficulty in the process of improving the satisfaction score. Finally, we find out target customers on the basis of satisfying degree and model utility at the same time. According to the influencing factors of each brand and the regression equation of each brand obtained in the previous article, the target customers are statistically analyzed, and finally a mining plan for customers of each brand can be given.