Search results for query: *

Different thresholds will offer different trade-offs between errors and throughput: at one end, errors are less frequent but sometimes no outcome is chosen; at the other end, some outcome is always chosen, but errors are more frequent. You will need to decide what trade-off best suits your problem.

Have you checked the documentation for the meaning of this error, or consulted the Weka Web page (http://www.cs.waikato.ac.nz/ml/weka/)?

If I have two series of numbers, series A contains either 1s or 0s, depending on if a patient took a pill or not. Series B contains random numbers. All of the series B numbers that coincide with the patient taking a pill have an average of 100, whereas those that coincide with NOT taking a pill...

Methodologies are largely checklists to help avoid overlooking anything. I don't think that one really presents a substantial advantage over the next. Personally, I use my own process, which I hone over time.

I wonder what experiences people here have had with open source data mining tools (Weka, Yale / RapidMiner, Orange, etc.)?

In most cases, the most predictive in data regarding delinquent customers likelihood to pay will be their activity with the loan product (purchasing and payment activity on a credit card, etc.). Credit bureau data is also popular, and I know some people have had success with demographic data...

Try the FAQ section here for the item titled "Where can I find more information on data mining?" -Will Dwinnell http://matlabdatamining.blogspot.com/

I work for the collections department of a bank, building predictive models of customer behavior. Out data is stored in Oracle, which I retrieve to a PC for analysis and model development in MATLAB. Did you have more specific questions?

Linked below is another paper on the subject of data mining versus statistics, "Data Mining and Statistics: What's the Connection?", by Friedman: http://www-stat.stanford.edu/~jhf/ftp/dm-stat.pdf -Will

I suppose association rule analysis (also called "market basket analysis") might work. You can find a list of commercial and free tools which perform such analysis at: http://www.kdnuggets.com/software/associations.html If, however, you know how the groups will be defined (model, color, A/C...

In general, for train/test splitting, I try to stratify as much as possible within reason, and yes, I do stratify on the dependent variable. "Within reason" means: 1. I worry most about variables believed to be important, and 2. individual stratification cells should not become too small. You...

Yes, try the 'pchip' or 'spline' functions in MATLAB.

That depends on what one is trying to accomplish. Why are you stratifying the data?

Readers here may be interested in my article,Family Recipe For Neural Networks, which was posted to the Data Mining and Predictive Analytics Web log: http://abbottanalytics.blogspot.com/2006/11/family-recipe-for-neural-networks.html#links I hope this is helpful!

There is a post on the Data Mining and Predictive Analytics Web log, Free And Inexpensive Data Mining Software, which may be of interest: http://abbottanalytics.blogspot.com/2006/11/free-and-inexpensive-data-mining.html Note the discussion which follows in the Comments section, as well.

There is an excellent set of data mining tutorial presentations by Andrew Moore at: http://www.autonlab.org/tutorials/ -Will Dwinnell http://matlabdatamining.blogspot.com/

Numerosity" means the number of distinct values in data. Some data miners will try to reduce this number for individual variables, either to compress the data set or to smooth the data. You can read more about this in "Predictive Data Mining" by Weiss and Indurkhya.

Since late last year, I have been writing a Web Log, called "Data Mining in MATLAB", which may be of interest to readers here. It is located at: http://matlabdatamining.blogspot.com/ -Will Dwinnell http://will.dwinnell.com

I moved away from commercial data mining tools several years ago in favor of writing my own code (in MATLAB). I was wondering: Do people here use commercial (or even free) tools, or do you build your own analysis software? -Will Dwinnell http://will.dwinnell.com

Registration for KDD 2006, which is in Philadelphia this year gets more expensive in just a few days: http://www.acm.org/sigs/sigkdd/kdd2006/ -Will Dwinnell http://will.dwinnell.com