I want to construct a Bayesian network from an incomplete data set. I want to know that while learning the structure of BN from data, which values should be set for the missed values in data set. I use WEKA for structure learning.
Do you have attribute-value pairs in your training/test set?
If so, the following steps are used to remove the errors due to the
missing attributes or the missing values of the training data set.
Step 1(Missing attributes): In this case, the missing attribute is added and its value is obtained based on step 2.
Step 2 (Missing values): the abundant value in the related class is selected for symbolic missing value (e.g. true/false) and the average of the values in the related class is calculated and set to the missing numeric values.
you can also check it out:
http://faculty.mccombs.utexas.edu/maytal.saar-tsechansky/ResearchPapers/saar-tsechansky07a.pdf
In my limited knowledge, one does not have to specify any thing for missing observations. One can fill random values in the cpt tables of these missing variables. While training, usually the network goes through two pass algorithm and assigns appropriate cpt values for these missing variables.
Tutorial of Bayesian Network toolbox might help one to figure out through examples. Link added
https://code.google.com/p/bnt/
- Badges
- Science topic
- Similar topics
- Psychology
- Cognitive Science
- Mind