Application Of Data Mining Technology In The Sale Of A Company's Notebook Computer

With the increasingly fierce competition in the industry,consumer market segmentation and lower information costs, modern maxketing occur great changes. Database marketing which is based on consumer adapted to this trend, gradually formed and developed rapidly. Database marketing is formed by the combination of database technology and marketing. Using database marketing technology, companies can be more accurate to find the target population, recommend appropriate products to customers at the appropriate time and improve the probability of customers to buy products. Database marketing technology not only can play a great role in the process of acquiring new customers but also can be applied to retain old customers and consolidate the market share. In this article, we briefly introduce the background knowledge about database marketing and explain the development process of database marketing.In the database marketing technology, establishing a reasonable and effective model is a very important part of marketing. Only when the model is reasonable and effective, the enterprise can more accurately find the target population, and then improve the profits and market share. In the process of establishing the model, data mining technology can play a great role. Data mining technology uses complex statistical modeling techniques to analyze information contained in massive data, and find the hidden rules. This paper briefly introduces the basic concepts of data mining and several common data mining techniques. The data mining process in the business field is also described in the article.The case study of this paper is based on a company's notebook computer sales data. We try to figure out whether customers that have bought notebook computer in the three years before December 2014 will buy the product again in the next three months. The main content of this paper are as follows:(1) A response model based on Logistic regression model is established. According to the third chapter of the paper, we introduce a series of processes to build the model, and use SAS software to build Logistic regression model. After establishing the model, the model fitting result is assessed by the model prediction accuracy and ROC curve and the robustness of model is also assessed by Index chart.(2) Based on variables contained in logistic regression model, BP neural network model, decision tree model, generalized additive model are established After the establishment of the several models, model prediction accuracy and area under the ROC curve are used to compare the fitting effect of several models. The robustness and com-plexity of models are also considered.(3) According to the results of several models, we pick customers that have higher probability of buying again and provide the final target population for the enterprise marketing activities.