2012 Raphael Saulus DBI 226 Data Mining for
© 2012 Raphael Saulus
DBI 226 Data Mining for Fun & Profit Russ Blake Principle Architect, Runge Limited Kevin Clarke Manager Software Development EB Games Australia, NZ Darien Nagle Solution Specialist Application Platform Microsoft Corporation
or…
or… …would you buy a Performance Monitor from this dude?
Explains why things fall
Explains why things fall 7 times more accurate
Runge Ltd Mine Planning Consultancy Planning, Scheduling and Forecasting
forecasting
Public Domain (Wikipedia Commons)
Public Domain (Wikipedia Commons)
The future will be like the past… © 2011 Microsoft Corporation
The future will be like the past… because… © 2011 Microsoft Corporation
The future will be like the past… because… in the past… © 2011 Microsoft Corporation
The future will be like the past… because… in the past… © 2011 Microsoft Corporation the future was like the past! -- Gerald M. Weinberg, An Introduction to General Systems Thinking
the top technology
the top technology
patterns
patterns make predictions © 2012 http: //www. holdemreview. com
patterns make predictions © 2012 http: //www. holdemreview. com
patterns make predictions © 2012 http: //www. holdemreview. com
patterns make predictions © 2012 http: //www. holdemreview. com A lot
Data Mining does not need a Cube!
SSAS ≠ Cube
SSAS ≠ Cube © blog. vi. Xra. org
SSAS ≠ Cube © recultured. com
Cube: © 2012 only. HDwall. Papers. com
Cube: Pretty high barrier to entry © 2012 only. HDwall. Papers. com
Cube: Pretty high barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube
Cube: Pretty high barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX
Cube: Pretty high barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses
Cube: Pretty high barrier to entry Data Mining: © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses © 2012 Microsoft Corporation
Cube: Pretty high barrier to entry Data Mining: Pretty low barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses © 2012 Microsoft Corporation
Cube: Pretty high barrier to entry Data Mining: Pretty low barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses © 2012 Microsoft Corporation Data Mining: Build Structure
Cube: Pretty high barrier to entry Data Mining: Pretty low barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses © 2012 Microsoft Corporation Data Mining: Build Structure Configure Model
Cube: Pretty high barrier to entry Data Mining: Pretty low barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses © 2012 Microsoft Corporation Data Mining: Build Structure Configure Model Make Predictions
Cube: Pretty high barrier to entry Data Mining: Pretty low barrier to entry © 2012 only. HDwall. Papers. com Dimensional Modelling: Build a Cube Learn MDX Construct Analyses …of the PAST © 2012 Microsoft Corporation Data Mining: Build Structure Configure Model Make Predictions …about the Future
Excel Data Mining Add-in Can it really be this easy?
© 2011 xkcd. com
© 2011 xkcd. com
© 2011 xkcd. com
The Black Swan The Impact of the Highly Improbable Nassim Nicholas Taleb
The Black Swan The Impact of the Highly Improbable Nassim Nicholas Taleb Central Thesis: All significant events are unpredictable!
http: //msdn. microsoft. com/enus/library/dd 776389%28 v=SQL. 100%29. aspx
Demo Data Mining Designer Your one-stop-shop for data mining
Public Domain (Wikipedia Commons)
Correlation Tree Node
Correlation Tree Node
P(Trousers | Girl) = 20 / 40
P(Trousers | Girl) = 20 / 40 P(Girl) = 40 / 100
P(Trousers | Girl) = 20 / 40 P(Girl) = 40 / 100 P(Trousers) = 80 / 100
P(Trousers | Girl) = 20 / 40 P(Girl) = 40 / 100 P(Trousers) = 80 / 100 P(Trousers | Girl) P(Trousers)
P(Trousers | Girl) = 20 / 40 P(Girl) = 40 / 100 P(Trousers) = 80 / 100 P(Trousers | Girl) P(Girl) (20 / 40) (40 / 100) (80 / 100) P(Trousers)
2 Weight Loc W W 3 Weight Sex W W W Weight Age Input Neurons W W Buy No W W W Hidden Neurons Output Neurons
2 Weight Loc W W 3 Weight Sex W W W Weight Age W Multilayer Perceptron Network No W W Input Neurons • W Buy W Hidden Neurons Output Neurons
2 Weight Loc W W 3 Weight Sex W W W Weight Age W Multilayer Perceptron Network No W W Input Neurons • W Buy W Output Neurons Hidden Neurons aka Back-Propagated Delta Rule Network
Lift Chart Operation Population Bike Buyers Random: 50% Ideal: 100% Targeted Data Mining: 85%
Handles dependencies
Handles shocks
Default PREDICTION_SMOOTHING = 0. 5
PREDICTION_SMOOTHING = 0. 2
http: //office. microsoft. com/en-us/excel-help/data-mining-add-ins-HA 010342915. aspx#_Toc 257717762 http: //msdn. microsoft. com/en-us/library/dd 776389(v=SQL. 100). aspx http: //www. sqlserverdatamining. com/ssdm/ http: //www. microsoft. com/downloads/en/details. aspx? displaylang=en&Family. ID=868662 dc-187 a-4 a 85 -b 611 -b 7 df 7 dc 909 fc http: //docs. media. bitpipe. com/io_25 x/io_25515/item_392177/Tableau_S_Mktg. Ltr_BI_IT. pdf http: //www. thearling. com/text/dmwhite. htm
Database and Business Intelligence Track: All Sessions Exam 467 (new) or 460 (upgrade) to MCSE Business Intelligence Find Me Later At the Friday 11 AM Meet and Greet
- Slides: 115