The Applications Of Database Marketing In Retail Industry

Compared to traditional marketing, database marketing has several advantages. For example, it can improve the effectiveness of marketing and can also enhance the relations between enterprise and customers, which is the lever of transformation from product-centric management system to customer-centric management system. Since the era of big data comes and the technology of data mining gets mature, database marketing plays an important role in tracking customers, understanding customers, predicting customers'demand, which brings infinite possibilities. The newly-developing marketing theory which is based on the data of customers and quantitative analysis has been widely used by many domestic and foreign enterprises.Quantile regression is a basic method of estimating objects' conditional quantile according to given factors which have an effect on objects. As the traditional regression method's (the least-square method) supplement and extension, quantile regression has many advantages such as equivariance, asymptoticy, robustness and so on. Besides, it can also measure the given factors' effect on objects at different points, which makes a more comprehensive explanation on objects.The data studied in this paper is from a foreign clothing retail company, which markets only by mails or sending lists of products by mails. We need to help them make targeted marketing. In the first stage, we determine the response models and customer segmentation models in database marketing. Firstly, by combining the analysis of statistical modeling with database marketing, we set up a customer segmentation model whose target variable is whether customers make a response to marketing, which is logistic regression. A customer segmentation model is established when company A needs to pick out the customers who meet their particular activity characteristics. The model can divide customers into several groups with different characteristics, which can help A pick out the customers meeting marketing characteristics. The two methods based on database marketing have different targets, which performs better when two methods are combined. In the second stage, we adopt stepwise regression and quantile regression with stepwise selection. After knowing customers' response, customers hope to know the number of orders generated by those who respond. Make a regression process on the dataset by stepwise regression and quantile regression with stepwise selection respectively. Finally, after examining, the quantile regression with stepwise selection performs better.