Hello Experts,
I’ve got a logical problem. I’m just a newbie in this theme.
I build a cube with Analysis Services. It works and I can brows the cube.
For Example this could be a possible result:
Color
RedBlueBlackYellow
Auto brand
Opel12
VW214
Mercedes1122
The count shows us how much cars were stolen.
Now I try to build a Data Mining structure that shows me the behavior between the auto brand and the color. (To build the report I used Excel 2007 and the Data Mining AddIn)
After build the Mining Structure the Colum name are 1, 2 or 4 (just the counts).
To search for a better solution I tried to build the report manual. (Now it looked like that)
Auto BrandCollorstolen
OpelRedno
OpelBlueyes
OpelBlackyes
OpelYellowno
VWRedyes
VWBlueyes
VWBlackno
VWYellowyes
MercedesRedyes
MercedesBlueyes
MercedesBlackyes
MercedesYellowyes
Now the Column names are correct and the Data Mining structure is more understandable.
But the second report you can’t build with a cube, because the Measures were just numeric.
My Questions:
- Where is my error?
- How do you build a Data Mining?
- Is there somewhere a tutorial could help me (I just founded tutorials that explains the different works of the different Data Mining Models)
- How would you build this problem in a Mining Structure?
Hope I don’t forget some information and hear you soon
Yours sincerely,
Alex
Please visit our portal site ssdm.com. You can find tutorials, tips and tricks as well as samples there.
As to your problem, it depends on what information you wish to analyze. For example, you can build a data mining model using some classification algorithms (such as decision trees and neural network). If you mark stolen as predictable, you can use your model to predict the probability that some specific type of car got stolen. You can also train a clustering model to assist in exploring your data.
Thanks,
