Structure Learning

We are continually adding to Netica's learning-from-data capability (structure, parameter, testing).  The first addition is the TAN Structure learning of version 5.00.  TAN structure learning is the automatic method for learning the link structure of a Bayes net from a case file.

How to:   First, select a target node that you want to diagnose/predict.  Next, choose Cases Learn Learn TAN Structure. Netica will automatically determine the appropriate link structure.    

As noted in the topic learning the link structure, there is no problem in having a great many links leaving a node, and since Netica will do Bayesian inference on the results, it is okay for links to go in either direction.  That is why to classify, predict or diagnosis a particular variable with the best accuracy, you want to capture its relation with as many of the other variables as possible, so you put many links leaving that variable.

Further documentation will be available shortly; in the meantime, you can read Friedman, Nir et al 1997 to discover how Netica does TAN learning (free version available online or by e-mailing us).

If you have some data sets that you would like us to use  throughout our research we would be happy to use them and send you all the results.