Data Mining Software tools - Critical Appraisals

[nchapman]

Does anyone know where I can get information (white papers etc) giving crictical independent appraisals of Data Mining Software tools. I am putting together a paper for my company, in which I would like to put forward some suggestions on what we should use.

For info. We are using an Oracle 8i database where our warehouse resides (will go to 9i soon). We are currently building some data marts, and envisage these will be used by business users, for OLAP and multi-dimensional analysis. Further down the line we would then like to use a Data Mining tool, in order to extract hidden patterns etc about our customers. We currently have SAS, SPSS and Clementine which the users currently only really use for data manipulation (a lot of which could be done on the database), and basic OLAP stuff (ie. describing historical data).

 

[JohnnyDarko]

Try the site

http://www.dwreview.com

If you are doing generic data mining - most of the data mining apps mentioned above will suffice. Choose one that your team members will already have some familarity with - this is usually the problem area.

Hope this helpw,

JD

 

[bkj123]

I agree with JD. SAS and SPSS/Clementine have good (very good in my opinion) products and even more importantly, since your users are familiar with them and have accepted them, you are setting yourself up for a successful implementation.

As far as reviews, JD's mention of DMREVIEW is always good or you could go with a think tank, etc. Sometimes it is hard to get the facts especially those relevant to your situation.

I would suggest that you nail down your scope/objective (including delivery time frame, budget, and support needs) of the project. Then, contact SAS, SPSS, and possiblly another vendor or 2 (e.g. Unica, Oracle) for a proposal. You could even throw a quick evaluation template together to compare each (total implementation cost, support, ease of use, which requirements don't they meet, etc.).

Think the question is not so much which vendor but which solution from that vendor? For example, if you are looking for a "data mining solution" from SAS...
- do we stick with the modeling PROCS in base SAS or do we go with Enterprise Miner (EM)?
- if we go with EM, do we run it from PC SAS, Unix, or an emulator, or.... (this can drives costs)
- do we keep using SPSS?
- Should we use oracle for all warehousing, data management, and manipulation or should we use SAS or a mix.

You get the idea. Hope this helps. Thanks - Brian