Since you are posting this in the Data Mining area I will make the assumption that you want an answer from that perspective. The answer depends on several issues.
- size and complexity of the data
- is it in a data dbase
- do you need the data mining tool to make the data "data mining ready"
- where do you want to score your results
SAS is good for raw data handeling But SAS Enterprise Miner has the easy to use mining tools. However SAS is probably the most popular
SPSS escpecially with Clementine is good
KXEN is a great system especiall if you prefer the process to be a "Black Box" it is fast and hadles lare data and has many ways to export the Scoring algorithms. Its weakness is in preparing the data for mining.
CART, MARS, etc
Statistica, R, Splus
I have handled small problems with the last two, medium problems with the first 2 and large problems with KXEN.
Usually your areas of pain in the data prep and production scoring sections of any data mining project.
I have also seen Weka, Teraminer and Fair-Isaac's Model Builder used for projects.
IMO Statistica is as good as SPSS and super cheap. I got my license for $125 through a grad school class. If you don't want to pay the full fee, and aren't enrolled, some grad classes are open to the public as part of an extension program, while others offer educational prices at their student store.
I am mainly using SAS Enterprise Miner to handle large marketing data. If customers need SAS/STAT as the final product, I will use SAS to do most of work, but SAS Enterprise Miner is a good tool to compare my final result.
I am currently looking for marketing data analysis job, if anyone has information, please leave me a message. Thanks.
I will toss my hat in for Unica. I have had great results with Unica for large-scale mining (million records or more). Smaller projects I fall back to SPSS. I have also heard good things about Clementine, although I believe SPSS is re-branding this under another name.