DAT332

Size: px

Start display at page:

Download "DAT332"

所撒
5 years ago
Views:

1 议程数据挖掘简介 Microsoft SQL Server Server 2005数据挖掘概要 2005数据挖掘概要介绍利用数据发掘方式使Business 利用数据发掘方式使Business Intelligence BI BI 变得更加智能用数据挖掘让BI 更智能化用数据挖掘让BI更智能化唐朝晖高级项目经理 SQL Server商务智能 Server商务智能 Microsoft公司 Microsoft公司智能多维数据库高级ETL 高级ETL 智能问题与解答数据挖掘能够完成哪些工作 Integrate 数据抽取,集中数据整合,转换 Analyze 数据维度模型数据处理数据挖掘 Report 流览数据建立模型预测未来数据表现数据发布查询多维数据库数据挖掘是什么为什么怎么做 SQL Server 2005 数据挖掘功能数据挖掘功能完整的算法集合决策树集群关联贝叶斯算法神经网络序列群集时间序列线性回归逻辑回归丰富的工具向导程序编辑器浏览器智能化多维数据集简单但非常强大的API XMLA和简单但非常强大的API XMLA和DMX DMX 与其它BI 技术的深度集成与其它BI技术的深度集成 DAT315 1

2 OLAP 挖掘应用情境 OLAP 数据挖掘多维数据集一种语义完善且具有预先计算好的集合值的数据库 OLAP 数据挖掘应用情景市场篮分析客户聚类客户分类销售预测 OLAP 挖掘模型与关系型相同灵活快速的建模挖掘模型客户 OLAP 数据挖掘挖掘模型客户挖掘维度挖掘 OLAP 多维数据集 AMO 对象模型概要介绍 AMO 代码 Server Database Cube CubeDimension MeasureGroup Mining Structure Mining Model Cube Attribute Measure MiningStructure Column MiningModel Column MiningModel Algorithm Algorithm Parameter MiningStructure myminingstructure = mydatabase.miningstructures.add("customersegement","customersegement"); ScalarMiningStructureColumn keystructurecolumn = Utilities.CreateMining MiningModel myminingmodel = myminingstructure.createminingmodel(true, true, "CustomerSegment"); //Add algorithm parameters DAT315 2

3 AMO 代码 1/4 AMO 代码 2/4 Cube mycube //Connect = mydatabase.cubes["foodmart to database 2000"]; MiningStructure Server myminingstructure myserver = mydatabase.miningstructures.add("customersegement","customersegement"); = new Server(); ScalarMiningStructureColumn ); keystructurecolumn = Utilities.CreateMining CubeDimension mydimension = //Add Store mycube.dimensions["customer"]; Sales as an attribute in mining structure MiningModel //Create myminingmodel = myminingstructure.createminingmodel(true, a new mining structure true, "CustomerSegment"); //Add algorithm MiningStructure parameters myminingstructure = mersegment"); myserver.connect("datasource=localhost;catalog=foodmart" mydatabase.miningstructures.add("customersegment","custo myminingstructure.source = new CubeDimensionBinding(".", mycube.id, mydimension.id); MiningStructure myminingstructure = mydatabase.miningstructures.add("customersegement","customersegement"); myminingstructure.source CubeAttribute = new CubeDimensionBinding(".", totalchildren mycube.id, = mydimension.id); ScalarMiningStructureColumn keystructurecolumn = Utilities.CreateMining MiningModel myminingmodel = myminingstructure.createminingmodel(true, true, "CustomerSegment"); //Add algorithm parameters AlgorithmParameter Sales"]; parm1 = new AlgorithmParameter("MODELLING_CARDINALITY", DINALITY", 1); Column(storeSales); mycube.dimensions["customer"].attributes[ TotalChildren"]; Column(totalChildren, false); Measure storesales = mycube.measuregroups[0].measures["store ScalarMiningStructureColumn storesalesstructurecolumn = AMO 代码 3/4 AMO 代码 4/4 Cube mycube //Create = mydatabase.cubes["foodmart a mining model 2000"]; from the structure MiningStructure MiningModel myminingstructure myminingmodel = mydatabase.miningstructures.add("customersegement","customersegement"); = CubeAttribute myminingmodel.algorithm customerkey = mycube.dimensions["customer"].attrib = ScalarMiningStructureColumn keystructurecolumn = Utilities.CreateMining MiningModelAlgorithms.MicrosoftCluster //Add algorithm StructureColumn(totalChildren, parametersfalse); AlgorithmParameter parm1 = new //Add Store AlgorithmParameter("MODELLING_CARDINALITY", Sales as an attribute in mining structure 1); ScalarMiningStructureColumn storesalesstructurecolumn = Utilities.CreateMining MiningModel myminingmodel = myminingstructure.createminingmodel(true, true, "CustomerSegment"); //Add algorithm //Process parameters the model //Process model myminingstructure.createminingmodel(true, "CustomerSegment"); Exception sre) Database //Create mydatabase = myserver.databases["foodmart"]; column binding to the cube attributes public static ScalarMiningStructureColumn MiningStructure myminingstructure = mydatabase.miningstructures.add("customersegement","customersegement"); //Create iskey) a key ScalarMiningStructureColumn keystructurecolumn = Utilities.CreateMining ScalarMiningStructureColumn column = new ScalarMiningStructureColumn(); column.name = attribute.attribute.name; Measure storesales column.iskey = mycube.measuregroups[0].measures["store = iskey; Sales"]; column.source = new MiningModel CubeAttributeBinding(attribute.ParentCube.ID, myminingmodel = myminingstructure.createminingmodel(true, true, "CustomerSegment"); //Add algorithm parameters AttributeBindingType.Name); column.type = CreateMiningStructureColumn(CubeAttribute attribute, bool ((CubeDimension)attribute.Parent).ID, attribute.attribute.id, MiningStructureColumnTypes.GetColumnType(attribute.Attribute.Na mecolumn.datatype); return column; 数据挖掘与集成服务高级 ETL 数据挖掘任务处理任务预测任务数据挖掘转换预测转换模型培训转换文本挖掘模糊查询数据挖掘准备清理采样 DAT315 3

4 利用集群模型检测异常数据收入利用集群模型检测异常数据收入? 在已知模型内的可能性?? 年龄? 年龄 PredictCaseLikelihood() Select Customer_Id, PredictCaseLikelihood() From MySegmentationModel Prediction Join OPENQUERY(mydatasource, Select Age, Income From Customers ) As t On MySegmentationModel.Age = t.age And MySegmentationModel.Income Select Top 10 Customer_Id = t.income From MySegmentationModel Prediction Join OPENQUERY(mydatasource, Select Age, Income From Customers ) As t On MySegmentationModel.Age = t.age And MySegmentationModel.Income = t.income Order by PredictCaseLikelihood() ASC 用数据挖掘技术进行数据清理集成服务与数据挖掘演示服务集成生成智能以 DMX 作为数据源设计环境中的预测查询生成器以作为数据挖掘预测结果展现方式利用预测分析方式生成由数据驱动的预测查询内容查询参数化查询 DAT315 4

5 数据挖掘 DMX DMX ADOMD.Net ADOMD.Net 服务器对象模型服务器对象模型系统系统用户用户 Microsoft Analysis Services 9.0 分析服务数据挖掘 // This function returns the list of mining models in the database public DataTable GetMiningModels() DataTable dt = new DataTable(); DataColumn dc = dt.columns.add("model Name", typeof(string)); dc = dt.columns.add("algorithm Name", typeof(string)); dc = dt.columns.add("description", typeof(string)); for (int i= 0; i<context.miningmodels.count; i++) DataRow dr = dt.newrow(); dr[0] =Context.MiningModels[i].Name; dr[1]= Context.MiningModels[i].Algorithm; dr[2]= Context.MiningModels[i].Description; dt.rows.add(dr); return dt ; 问题与解答智能与数据挖掘演示问题与解答利用利用提问提问按钮以文字方式提交问题按钮以文字方式提交问题请不要忘记填写调查表请不要忘记填写调查表如需了解即将进行或以往的现场Web 广播请如需了解即将进行或以往的现场Web广播请访问访问如需获取Web 广播内容提要请发送电子邮件如需获取Web广播内容提要请发送电子邮件至 = 目前Web 广播需通过Microsoft Microsoft Office Live 目前Web广播需通过 Meeting收听获取为期 14天的免费使用版本天的免费使用版本 Meeting收听获取为期 DAT315 5

6 DAT315 6

Microsoft PowerPoint - DAT_SQL2005DM_ZhaoTang.ppt

Microsoft PowerPoint - DAT_SQL2005DM_ZhaoTang.ppt SQL Server 2005 之数据挖掘提纲数据挖掘概述什么是数据挖掘? 典型商务问题 SQL Server 2005 系统中的数据挖掘 Key messages DMX 演示 Q&A 新特征介绍有关工具分析引擎 BI 集成什么是数据挖掘? 数据挖掘 ( Data Mining), 又称信息发掘 (Knowledge Discovery), 是用自动或半自动化的方法在数据中找到潜在的, 有价值的信息和规则.