BayesFusion Support Forum

By changing the definition of the child node. The conditional probability distributions describe precisely the impact that the parents have on it. In your model, the conditional probability distributions (columns in your CPT) are all equal.

I hope this helps,

Marek

The guilty element of your model is the CPT in n10:

s0 0.5 0.5 0.5 0.5
s1 0.5 0.5 0.5 0.5

Please note that the probability distribution over n10 is not impacted by the parents.
Does this help?

Marek

Thank you for sending me your model. I have looked at it and saw that quite likely you mean the nodes n0, n6, n3, n4, n12 , n11 and n9 in your obfuscated model. Each of these nodes is defined by a CPT that consist of only zeros and ones. Zero and one are special numbers in probability theory. Once a...

Both parameters are accessible from the Learn New Network dialog. The upper right pop-up menu allows you for choosing the algorithm. Have you actually looked at GeNIe manual :-)? You can send me the data set by a private message on the Forum or by Email (marek@bayesfusion.com).
Cheers,

Marek

Hi Ellen, If the algorithm that you are using is the PC algorithm, it is possible that the structure that it discovers is too complex. There may be nodes that have many parents. In that case, the crash occurs when you go to the next step, i.e., learn parameters. The probability tables are huge and m...

A probability distribution that does not add up to one is impossible and wrong. GeNIe will not let you exit the dialog in which you enter such distribution. I don't know for sure, just guessing, but I think that you may want to represent a dummy probability distribution that will never be used in so...

Thank you for the additional description. The lognormalCDF() does make sense (I assume that this is what your theory dictates). I still think that if you want to make the two nodes parents of Node5, you should make sure that the discretization of both makes sense and then specify the CPT of Node5 co...

I will be happy to try to help. I have a hard time understanding the definition of the node "damage state". You are making several calls to the Uniform() function and this confuses me somewhat. Please note that as Uniform() generates a random number from the Uniform distribution, each time...

Absolutely! Please try Layout menu and then Graph Layout. There are two simple algorithms that do a reasonable job in rearranging nodes for readability.
I hope this helps,

Marek

Hi Salim,

The screen shots are not originating from GeNIe and are confusing. My original advice is still standing. You should change the uniform probabilities into the probability distributions that will be there when the node Risks has no effect on the children nodes.
Cheers,

Marek

Hi Salim, I will be happy to help you but need to understand better what you want. Would you be willing to create and post the simplest possible influence diagram that does what you want? In general, uniform probabilities do not sound like a good idea in this case. I would keep the probabilities tha...

Hi Maghnie, The silent crash has to do with the fact that we have not yet programmed this functionality for continuous nodes. We will fix this in one of the next releases. To make it work at the moment, please change the type of the continuous nodes from continuous/equation to discrete chance nodes ...

It seems to me that this is a data set that you can use to train a Bayesian network. An influence diagram would require explicit decisions and also a utility function. Your data set is small (34 records) and you cannot really learn anything useful from data alone. If you create a BN model based on (...

Probability of any evidence in a continuous node will be zero, so you cannot do that even theoretically. I suggest that you discretize your network first and then calculate P(E). You can make the interval around the evidence as small as you want but not empty/zero length.
I hope this helps,

Marek

The Bayesian Search-based algorithms (Bayesian Search and GTT) learn no arcs, PC and Naive Bayes-based algorithms (NB, TAN and ANB) show plenty of dependencies. With 12 data records, whatever you will learn in terms of structure will be completely unreliable. Naive Bayes may do best if one of your v...