Why do I get marginal edge probabilities >1?
Hi, I am trying to perform network reconstruction using MeasuredBlockState. I then collect the marginal edge probabilities which are stored as an EdgePropertyMap 'eprob'. Would it be possible to explain why some probabilities in 'eprob' are >1? My understanding was that eprob of edge (a,b) is interpreted as the /marginal posterior probability of edge(a,b) to exist in the inferred network/, in which case all instances of eprob should be <=1. Would it be possible to explain why this is not the case? Thank you in advance. -- Sent from: http://main-discussion-list-for-the-graph-tool-project.982480.n3.nabble.com/
Am 29.07.19 um 14:49 schrieb marisdem:
Hi,
I am trying to perform network reconstruction using MeasuredBlockState. I then collect the marginal edge probabilities which are stored as an EdgePropertyMap 'eprob'. Would it be possible to explain why some probabilities in 'eprob' are >1?
My understanding was that eprob of edge (a,b) is interpreted as the /marginal posterior probability of edge(a,b) to exist in the inferred network/, in which case all instances of eprob should be <=1. Would it be possible to explain why this is not the case?
Could you provide a minimal and self-contained example that shows the problem you are describing? (Note that get_edge_prob() actually returns the log-probability in the range [-inf, 0], not the probability in the range [0,1].) Best, Tiago -- Tiago de Paula Peixoto <tiago@skewed.de>
participants (2)
-
marisdem -
Tiago de Paula Peixoto