110 likes | 364 Views
Data mining using SQL SERVER 2005. My name: ZULFIQAR SYED Holds BSEE from Illinois Institute of Technology. MCP in ASP.net (C#) SQL SERVER, ASP.NET, C#, DATA MINING, ANALYSIS SERVICES. CONTACT: DATAGIG@GMAIL.COM HTTP://ZULFIQAR.TYPEPAD.COM. Prerequisites for data mining. SQL SERVER
E N D
Data mining using SQL SERVER 2005 My name: ZULFIQAR SYED • Holds BSEE from Illinois Institute of Technology. • MCP in ASP.net (C#) • SQL SERVER, ASP.NET, C#, DATA MINING, ANALYSIS SERVICES. • CONTACT: • DATAGIG@GMAIL.COM • HTTP://ZULFIQAR.TYPEPAD.COM
Prerequisites for data mining • SQL SERVER • T-SQL
Business Problem • How to recommend movies based on customer demographics. • How to recommend other movies only based on movies already in shopping basket.
Demonstration • Simple DMX query • Structure, • models • prediction Query • Nested DMX query • Structure • Nested models • nested query
Demonstration recap. • Created/Trained/Queried • simple case model. • Nested case model. • Predict based on demographics. • Predict based on already bought items.
Creating Structures/Models • Create Structure • Define key column. (normally primary key) • Define other influencing columns. • Define Nested Table • Define key • (NOT primary key) • Depending on context • Add one or more models • Indicate prediction column(s). • Algorithm • Parameters (Optimization)
Structure/Model columns. • Create Structure (Similar to creating OLTP tables) • columns • data types • Long • Double • Text • Date • Table (for nested table) • Content Types • Continuous • discrete • Add model(s) to structure • column(s) to predict. • Input (default) • Predict • Predict_only
Prediction Query Basics • Prediction Query basics (Similar to OLTP select) • (psuedo code) • Select <column list> From <mymodel> Join <myinput table> On <column list> Where <clause> • Make cross services call • OpenQuery (preferred, only specify datasource object) • OpenRowSet (expose credentials) • Join • Prediction • Natural Prediction Join
Algorithms • Decision Trees • Nodes • Split • Parameters • Nodes • Association Rules • Item Sets • Importance • Exist
Model Training • Similar to Populating OLTP table. • Insert into model, select query • Shape operator for nested tables. • Skip operator for irrelevant primary key in nested table.
Q and A • Books • Data mining Techniques (Berry/Linoff) • Data mining with Sql Server 2005. (Tang/MacLennan) • Please fill out the evaluation form. • NAME: ZULFIQAR SYED • SESSION: Relating SQL SERVER 2005 DATA MINING to Business Issues • My contact information: • DATAGIG@GMAIL.COM • Web log: HTTP://ZULFIQAR.TYPEPAD.COM These slides will be posted on my web log.