SAS 数据挖掘实例-广告投放问题eng.docVIP

SAS Tutorial: Enterprise Miner Case description Suppose that you work for a mail order enterprise that sends out a catalog of furnishings and housewares each month. As part of an upcoming sales campaign, you want to distribute a special catalog that is devoted to fine dining and contains kitchenware, dishes, and flatware. Its too expensive to send this catalog to all of your customers, so you need to target those most likely to buy. You do this by developing a targeting model and then using it to produce a new mailing list. You have an extensive record of customer purchases. The data includes variables that indicate whether customers bought kitchenware, dishes, or flatware in the past two years. This purchase history has been used to create the CUSTDET1 data set, which contains 49 variables with the following labels: PurchaseDollars SpentYearly IncomeHome ValueOrder FrequencyRecencyMarriedName PrefixAgeSexTelemarket Ind.Rents ApartmentOccupied 1 YearDomestic ProductApparel PurchaseLeisure ProductLuxury Items Kitchen ProductDishes PurchaseFlatware PurchaseTotal Dining (kitch+dish+flat)Promo: 1-7 MonthsPromo: 8-13 Months$ Value per MailingCountry CodeTotal ReturnsMens ApparelHome FurnitureLamps PurchaseLinens PurchaseBlankets PurchaseTowels PurchaseOutdoor ProductCoats Purchase Ladies CoatsLadies ApparelHis/Her ApparelJewelry PurchaseDate 1st OrderTelemarket OrderAccount NumberState CodeRaceHeating TypeNumber of CarsNumber of KidsTravel TimeEducation LevelJob Category Enterprise Miner Based on SAS software, Enterprise Miner combines the data mining process with graphical ease of use. It delivers a broad range of predictive and descriptive models that you can apply, test, and compare to determine the best fit for the data. Most of this is done by manipulating icons: you connect nodes in a graphical workspace, adjust settings, and run the workflow. Heres an example of the workflow: Nodes are a key concept in Enterprise