0

I have a dataset that contains integral values that are discrete with a constant interval. I am trying to learn a Bayesian Network. I tried to convert these values into float and used the discretize function from the bnlearn package on R. I am using the interval method, but I need a large number of breaks to be specified and running into errors.

Here is a small sample of the dataset that I am using to learn the Bayesian Network.

X Y P1 P2 P3 P4 P5 P6 P7 P8 P9 0,20,123,125,117,122,127,112,123,125,116 0,21,120,136,129,116,145,125,118,136,125 0,22,122,126,129,116,121,129,117,122,127 0,23,130,124,127,126,119,123,126,121,124 0,27,126,122,124,118,120,128,122,122,125

Could someone please tell me how I can retain the individual values of the data while still being able to learn the Bayesian Network.

Thanks Rochan

Rochan Avlur
  • 67
  • 1
  • 8
  • This is currently quite broad. Please narrow it down to what package(s) you are using and what you've tried. A good bonus would be to include a reproducible example in a form where we can copy/paste it into our R console and start working right away. [Here are a few tips on how to do just that](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example). – Roman Luštrik Jan 24 '17 at 09:07
  • Hard to say without knowing what the data is. Are the variables really discrete, or are they just measured as integers (not integrals i assume): could you run them through the gaussian learner . If you specify a large number of breaks, then your probability tables could end up very large so you will need *a lot* of observations for this (and more breaks does not necessarily mean an improved performance) – user20650 Jan 24 '17 at 13:47

0 Answers0