05_决策树算法案例_高校学生用户识别模型.ppt

* * . 7:15 – buffet opens 7:45 – Rick starts program 1 – Welcome -= 5 min 2 – Recognize value of partners 3 – Intro TD LT available for QA, and Scott Collins (others?) 4 – Intro key speakers 5 – Personally introduce Mike – Mike speaks 10-15 min 6 – Mike to intro Randy – Randy speaks 30 min - feature 7 – Rick – wrap-up and QA Rick to provide a few comments – maybe a few slides on who partners are – SI and ISVs who are attending. Rick to introduce any other members of the LT who are in the audience. Darryl-yes; Bob-conflict right before; Scott Gnau-? All LT people at front table. – Executives took the time to be there and sit in the room with them. Acknowledge all partners that are there. Might not be necessary to show logos. Other slides? Rick probably doesn’t need any slides – maybe just Agenda slide. Set the tone and how critical and important they are to building an overall TD ecosystem. How part of a whole ecosystem and application development environment that pulls customers to TD from solutions. Be part of our success and growth of company; talk about how together we should be innovating what we do around AEI, future of EDW, tie into that as a theme into Randy. About 130 people last year. (133 confirmed; 35 were TD) Expect about the same as last year. About 15 TD people – will accompany partners. Teradata Confidential * * 高校学生用户识别模型_模型分析报告 Teradata SC Jul, 2010 目录 2 高校学生样本分析 3 模型构建 分析目标及统计口径 1 4 模型结果 5 模型评估、部署与维护 统计口径及分析目标 分析目标 根据学生群体的语音通话、行为轨迹、假日行为、新业务使用特征、交往圈信息等特点，建立相应的学生客户识别模型，进行针对性营销与服务 。 统计口径 分析用户群 移动公司提供的一批归属于贵阳市贵州大学、贵阳学院、贵阳医学院三所高校的前置营销号码。 高校V网用户。 有在高校区域基站通过话的贵阳移动用户。 时间窗口 201003-201005 模型目标 构建一个分类模型，将在高校区域基站有通话的用户分为学生和非学生两类。 目录 2 高校学生样本分析 3 模型构建 分析目标及统计口径 1 4 模型结果 5 模型评估、部署与维护 高校学生样本分析 目标： 提高学生样本纯度 结论： 对于第一批前置营销用户，按照如下规则筛选： 201005月底用户状态正常 2007-2009年8/9月份入网 区县代码与校园位置一致 201005有在校园区域基站通话 对于三所学校校园V网的用户，按照如下规则筛选： 201005月底用户状态正常后入网 201005有校园通话。 高校学生样本分