Probability of 0.5

The front end.
Post Reply
tloy1966
Posts: 20
Joined: Thu May 12, 2011 2:40 am

Probability of 0.5

Post by tloy1966 »

Dear All

In Genie ,i use following procedure to built bayesian network(BN)

1. built model
2. match model and data with EM algorithm to learn Conditional Probability Table(CPT)

then i get model with CPT,

but some probability value in CPT is 0.5

Is this because of no related data , so EM algorithm cant learning probability?

Kind Regards,
Wei
mark
Posts: 179
Joined: Tue Nov 27, 2007 4:02 pm

Re: Probability of 0.5

Post by mark »

There could be several reasons, but no data is the most likely one.
tloy1966
Posts: 20
Joined: Thu May 12, 2011 2:40 am

Re: Probability of 0.5

Post by tloy1966 »

Thank you for reply.

And the others reason are :?:
mark
Posts: 179
Joined: Tue Nov 27, 2007 4:02 pm

Re: Probability of 0.5

Post by mark »

The probability could be exactly 0.5 in the data or the distribution could be fixed.

Have you tried randomizing? That could break the 0.5 probability.
tloy1966
Posts: 20
Joined: Thu May 12, 2011 2:40 am

Re: Probability of 0.5

Post by tloy1966 »

mark wrote:The probability could be exactly 0.5 in the data or the distribution could be fixed.

Have you tried randomizing? That could break the 0.5 probability.
Thanks you for reply!!! :)

I didn't tried randomizing, in my data all variables are observed ,due to i use complete data when learing the conditional probability table.

so, i think that could be distribution fixed.
mark
Posts: 179
Joined: Tue Nov 27, 2007 4:02 pm

Re: Probability of 0.5

Post by mark »

Distributions are only fixed when you fix them yourself, which I suspect is not the case. Having complete data is not a sufficient condition to get rid of 0.5 probabilities. What you need is that for each family (child plus its parents) all parent configurations occur in the data. Especially when a node has many parents, this does not always happen.
Post Reply