Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

I have the joint distribution of the bayesian network defined as

$P(CERFD)=P(C)\times P(R)\times P(E|CR) \times P(F|E) \times P(D|F)$

and I am trying to calculate $P(D=d|C=c)$, below are my workings and I am stuck at the last step:

$P(D=d|C=c)= \frac{P(D=d,C=c)}{P(C=c)} $ $=\frac{\sum_{ERF}P(C=c)\times P(R)\times P(E|C=c,R) \times P(F|E) \times P(D=d|F)} {\sum_{DERF}P(C=c)\times P(R)\times P(E|C=c,R) \times P(F|E) \times P(D|F)}$

$=\frac{\sum_{ERF}P(C=c)\times P(R)\times P(E|C=c,R) \times P(F|E) \times P(D=d|F)} {\sum_{DERF}P(C=c)\times P(R)\times P(E|C=c,R) \times \sum_F P(F|E) \times \sum_D P(D|F)}$

$=\frac{\sum_{ERF}P(C=c)\times P(R)\times P(E|C=c,R) \times P(F|E) \times P(D=d|F)} {\sum_{ER}P(C=c)\times \sum_R P(R)\times \sum_E P(E|C=c,R) \times 1}$

$=\frac{\sum_{ERF}P(C=c)\times P(R)\times P(E|C=c,R) \times P(F|E) \times P(D=d|F)} {P(C=c)\times 1}$

I don't even know how to simplify the numerator at all... This is not homework, I am learning probabilistic graphical model from coursera and I'm testing my own understanding.

share|improve this question
    
Don't bother with your initial factorization into P(D=d,C=c)/P(C=c); you can leave the conditional statement in throughout the math. For the numerator, note that P(D=d|F) X P(F|E) = P(D=d|E). It's helpful also if you write out the full conditional statements, e.g., P(D=d|E,F) X P(F|E) = P(D=d|E) throughout; I think that lacking those full conditionals may be where your confusion comes from. –  jbowman Mar 30 '12 at 14:24
    
$\Pr(C=c)$ is a constant for given $c$ so you can cancel that from the numerator and denominator. –  Henry Mar 30 '12 at 14:37
    
Thank you for answering. Do you mind directing me to the proof of P(D=d|F) X P(F|E) = P(D=d|E) and P(D=d|E,F) X P(F|E) = P(D=d|E)? My understanding of conditional probabilities is too weak. –  King Mar 30 '12 at 21:53

1 Answer 1

From your given conditional probabilities, we can marginalize the intermediate terms in the following manner:

$p(e | c) = \sum_r p (e | c, r) p (r)$

$p(f | c) = \sum_e p(f | e) p(e | c)$

$p(d | c) = \sum_f p(d | f) p(f | c)$

Putting everything together, we end up with the following:

$P(d | c) = p(d|f) p(f | e) p (e | c, r) p (r) = \sum_f p(d|f) \sum_e p(f | e) \sum_r p (e | c, r) p (r)$

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.