1
$\begingroup$

I want to calculate the joint probabilities P(X,Y), where X and Y are both categorical variables with 7 and 6 categories each. Given the conditional probabilities P(X|Y), is it possible to calculate the joint probabilities P(X,Y)?

If it is not possible from just P(X|Y), would it be possible if I had both P(X|Y) and P(Y|X)? As I have a model for estimating the conditional probabilities, but not the joint probabilities. And how would this exactly be done? I learnt how to do this in the case that both X and Y are binary variables, from this question,

Calculating joint probability and covariance given conditional probabilities

but I am not quite sure how to extend this to the multinomial variables case. Thank you so much!!

$\endgroup$
  • 1
    $\begingroup$ I think you could extend that same problem with more categories. The second answer shows the method. $\endgroup$ – dankernler May 10 '18 at 11:56
1
$\begingroup$

If you have both conditional probabilities, you can find the joint using Brook's lemma. In your case you have $$ P(X=x,Y=y) = P(X=x_0,Y=y_0)\frac{P(X=x|Y=y_0)}{P(X=x_0|Y=y_0)}\frac{P(Y=y|X=x)}{P(Y=y_0|X=x)} $$ where $(x,y)$ are the points where you want to evaluate the probabilities, and $(x_0,y_0)$ is a set of possible values. Then, computing $P(X=x,Y=y)$ for all possible values and remembering that their sum must be 1, you can find the joint density.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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