Microsoft PowerPoint - DAT_SQL2005DM_ZhaoTang.ppt

Size: px

Start display at page:

Download "Microsoft PowerPoint - DAT_SQL2005DM_ZhaoTang.ppt"

牵送党
5 years ago
Views:

1 SQL Server 2005 之数据挖掘

2 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

3 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. 数据挖掘技术来源于数据库数据库, 统计和人工智能. 分析问题整合数据建立模型理解规则预测未来

4 典型的商业问题客户的信用风险是什么? 我的客户是哪些? 客户比较倾向于同时买哪些商品? 下个月我能销售多少 Pepsi 可乐? 我的商店的潜在客户会有多少? 在互联网上我的主要客户群又是哪些? 微软的股票走向如何?

7 SQL Server 2005 数据挖掘

8 微软的解决方案 SQL 2000 SQL Yukon 进入 DM 领域创立工业标准整合开发者建议推出产品步入市场 V1.0 版本中已包括两个高性能算法成为 DM 领域领导者面向软件开发人员完善行业标准完备的分析功能向企业级应用渗透

9 数据挖掘与 Microsoft 商务智能 Dev Tools Visual Studio.Net Excel OWC Map Point Data Analyzer Balanced Score Card Reporting Services Analysis Services OLAP & Data Mining Integration Services SQL Server Relational Engine Windows Server SharePoint Portal Server Management Tools Windows Client

10 Key Messages 嵌入式数据挖掘集成平台从 DM 到 LOB 都可嵌入应用 DM SQL 语言 -DMX XML/A 支持高端分析与 ISVs 联盟

11 Key Messages 嵌入式数据挖掘集成平台高端分析与 ISVs 联盟与关系数据库 (relational),olap,dts 和汇报 (reporting) 技术的紧密集成 SQL Server: BI 的基础平台

12 Key Messages 嵌入式数据挖掘集成平台高端分析主流的数据挖掘算法增强的可视化控件与 ISVs 联盟

13 Key Messages 嵌入式数据挖掘集成平台高端分析与 ISVs 联盟倾力关注于市场的拓展基于 DMX 和 XML/A 的 DM 工业标准的最终统一第三方算法 Plug-in

14 API: DMX CREATE MINING MODEL CreditRisk (CustID LONG LONG KEY, KEY, Gender TEXT TEXT DISCRETE, Income Income LONG LONG CONTINUOUS, Profession TEXT TEXT DISCRETE, Risk Risk TEXT TEXT DISCRETE PREDICT) USING USINGMicrosoft_Decision_Trees INSERT INTO INTO CreditRisk (CustId,, Gender, Income, Profession, Risk) Risk) Select Select CustomerID,, Gender, Income, Profession,Risk From From Customers Select SelectNewCustomers.CustomerID, CreditRisk.Risk, PredictProbability(CreditRisk) FROM FROMCreditRisk PREDICTION JOIN JOINNewCustomers ON ONCreditRisk.Gender=NewCustomer.Gender AND ANDCreditRisk.Income=NewCustomer.Income AND AND CreditRisk.Profession=NewCustomer.Profession

15 简单数据简单数据 IQ IQ Not Encouraged Not Encouraged Encouraged Encouraged Not Encouraged Not Encouraged Encouragement Encouragement Parent Parent Income Income Yes Yes Male Male 3 Yes Yes Female Female 2 No No Male Male 1 College College Plans Plans Gender Gender StudentID StudentID

16 复杂数据 Cust ID Age Marital Status IQ Favorite Movies Title Score 1 35 M 2 Star Wars 8 Toy Story 9 Terminator S 3 Star Wars 7 Braveheart 7 The Matrix M 2 Sixth Sense 9 Casablanca 10

17 Nested Input Data CustID Gender Male Male Female Male Female Marital Status Married Married Single Single Married Education Bachelors Bachelors Masters Some College Bachelors Home Ownership Rent Own Own Own Rent CustID Movie Lord of the Rings Matrix Star Trek Terminator Star Wars E.T Female Married Masters Rent Star Wars Sixth Sense A Beautiful Mind Hours Signs Moulin Rouge Nested Key (not foreign key) Die Hard Apocalypse Now

18 High Express Power with Nested Table CREATE MINING MODEL MovieAssociation ( [CustID]] LONG KEY, [Gender] TEXT DISCRETE, [Marital Status] TEXT DISCRETE, [Education] TEXT DISCRETE, [Home Ownership] TEXT DISCRETE, [Movies] TABLE PREDICT ( [Movie Name] TEXT KEY ) ) USING Microsoft_Association_Rules

19 完备集算法决策树聚类时间序列序列聚类关联 Naïve 贝叶斯神经网络

NET AMO Any App Any Platform, Any Device WAN XMLA Over TCP/IP XMLA Over HTTP

20 Integration Services Reporting Services C++ App VB App.Net App OLEDB (msolap90.dll) ADO/DSO ADOMD.NET AMO Any App Any Platform, Any Device WAN XMLA Over TCP/IP XMLA Over HTTP Analysis Server (msmdsrv.exe) OLAP Server ADOMD.NET Data Mining DM Interfaces (dmalgo.idl) Data.Net Stored Procedures Microsoft Algorithms Third Party Algorithms

21 演示

更多信息 DM URL www.sqlserverdatamining.com 新闻组 : Microsoft.

22 更多信息 DM URL 新闻组 : Microsoft.Public.SQLServer.Dat amining 更多 : KDNugget.com ACM SIGKDD ICDM

3.1 SQL Server 2005 Analysis Services Unified Dimension Model (UDM) 3 ( Ad-Hoc) SQL Server 2005 E - R T-SQL(Star Schema) (Data Mart) (ETL) ( Ora

C H A P T E R 3 SQL Server 2005 SQL Server 2000SQL Server SQL Server 20052005 3.1 SQL Server 2005 Analysis Services 3.1.1 Unified Dimension Model (UDM) 3 ( Ad-Hoc) SQL Server 2005 E - R T-SQL(Star Schema)