2009-BI-No07-Topic08-BI-Sample-DMX-SQLServer2005.docVIP

DMX实验vTargetMail（视图）和ProspectiveBuyer。表vTargetMail用于训练模型，表ProspectiveBuyer用于预测。 2.1挖掘用数据集市模式 2.2 vTargetMail（视图）模式 2.3 表ProspectiveBuyer模式 3. 实验步骤 （1）在SQL Server Management Studio 中 连接 AS服务器； （2）选定项目； （3）新建查询； （4）调试MDX。 4. 决策树分类分析 下面是一个应用DMX创建挖掘结构、挖掘模型，训练挖掘模型及进行预测的完整过程例子。 条件：在Besiness Intelligence Visual Studio中建立了数据源Adventure Works DW（与AdventureWorksDW连接）。 .1 模型创建 create mining model Target_mail ( CustomerKey long key, Age long Continuous, CommuteDistance Text Discrete, NumberCarsOwned long Discrete, NumberChildrenAtHome long Discrete, Region Text Discrete, TotalChildren long Discrete, YearlyIncome long continuous, BikeBuyer long Discrete predict ) using Microsoft_Decision_Trees --上述创建过程中，同时产生一个挖掘结构Target_mail_Structure和该结构化的一个挖掘模型Target_mail。 --也可以采用分步方式实现 --先定义结构Target_mail_Structure /* create mining structure Target_mail_Structure ( CustomerKey long key, Age long Continuous, CommuteDistance Text Discrete, NumberCarsOwned long Discrete, NumberChildrenAtHome long Discrete, Region Text Discrete, TotalChildren long Discrete, YearlyIncome long continuous, BikeBuyer long Discrete ) */ --再基于该结构构建一个挖掘模型 /* alter mining structure Target_mail_Structure add mining model Target_mail_1 (CustomerKey not null, --not null：建模标志 Age, CommuteDistance, NumberCarsOwned, NumberChildrenAtHome, Region, TotalChildren, YearlyIncome, BikeBuyer predict --predict预测子句 ) using Microsoft_Decision_Trees; */ 4.2 模型训练 Insert into Target_mail ( CustomerKey, Age, CommuteDistance, NumberCarsOwned, NumberChildrenAtHome, Region, TotalChildren, YearlyIncome, BikeBuyer ) openquery([Adventure Works DW], select --属性与上述对应 CustomerKey, Age, CommuteDistance, NumberCarsOwned, NumberChildrenAtHome, Region, TotalChildren, YearlyIncome, BikeBuyer from vTargetMail ) --查看模型内容，模型必须训练后才可以查看 select * from Target_mail.CONTENT select attribute_name, node_type,node_caption, [children_cardinality], node_description, marginal_rule from Target_mail.CONTENT order by node_caption 4.3 预测查询 --简单预测，单独预测 有一个用户：