Statistical Analysis Of Customer Segmentation Of Petroleum Sales Company

In recent years,the profits of the oil industry have declined year by year,and the loss of customer resources is serious.All the sales companies have been thinking about how to serve customers better.However,because of the diversification of customer needs,we can not carry out a "one size fits all" marketing strategy for all customers,and we should carry out accurate marketing according to the characteristics of different customers.The same customer group usually has similar consumption habits and preferences.We can divide customers into different groups and carry out corresponding marketing activities for different groups,which can not only save the cost of the company,but also make the effect significant.The key to this is the effective customer segmentation.The key of customer segmentation is to dig out the features hidden behind the data through a series of statistical analysis methods,and then classify them.Nowadays,there are many research on customer segmentation,not only the traditional RFM model,but also many models based on statistical algorithms,but the research on the oil sales industry is not much,and the customer's index dimension is relatively simple,which is not enough to fully reflect the customer's consumption habits.Therefore,this article has certain theoretical and practical significance.In the first chapter,I introduce the background and purpose of the article,and the domestic and foreign research status from two aspects:the customer segmentation method and the customer segmentation application in the petroleum sales industry.Then I introduce the framework and possible innovation of the article.In the second chapter,I introduce the concepts,principles and methods of customer segmentation theory.In the third chapter,I mainly introduce the statistical methods used in this paper,including cluster analysis,K-means algorithm and principal component analysis.I introduce the advantages and disadvantages of K-means algorithm in detail,and I have improved how to determine the number of classifications K and how to select the initial aggregation point.In the fourth chapter,after taking full consideration of the classification targets,business data and customer characteristics of the oil sales industry,based on the detailed data of their personal gas card customers,I set up the individual card customer segmentation indicators from many dimensions.These indexes not only have traditional Monetary,frequency,Recency,but also include a description of the time of refueling,the location of the fuel,the quality of the oil.They show customers' consumption habits more comprehensively.In the fifth chapter,I build the customer segmentation model.According to the customer's consumption ability(value attribute),I first divide the customer into three categories of high,medium and low value.Then,according to the consumption habit(behavior attribute),I further subdivide the customers of different values,use the principal component analysis method to compress the clustering index,and then use the improved K-means algorithm to cluster analysis.Finally,I divide the high value customers into nine categories,the value customers are divided into eleven categories,the low value customers are divided into two categories.In the end,I explain the different subdivision groups,give their own prediction for the professional attributes of the nine types of high value customers,and put forward the corresponding marketing service suggestions.In the sixth chapter,I summarize the research results in the whole paper,and illustrates the problems and shortcomings.