Prior distributions?
Hi everyone! Does anyone know which prior distributions does Dynamic Yield use for conversion rate and for Average Revenue Per User?
Does it uses noninformative priors or subjective priors? And in each case, how do they model them, with what parameters?
Thanks!

Hi Ezequiel,
for binary metrics like CR:
 Set a minimum P2BB value to declare a winner
 Prior distribution  Beta(1,1)
 Gather experiment data per variation, namely # of successes (S) and # of observations (N)
 Posterior distribution Beta (S+1,NS+1)
 Sample n observations from each posterior (MonteCarlo)
 Calculate P2BB by comparing samples
for nonbinary metrics like AOV
 Set a minimum P2BB value to declare a winner
 Prior distribution  Gamma(K,T)
 Gather experiment data per variation, namely # of purchases (N) and sum of all purchases values (S)
 Posterior distribution of 1/AOV is Gamma(K+N, T/(1+T*S)
 Sample n observations from each posterior (MonteCarlo)
 Transform each sampled value to 1/(sampled value)
 Calculate P2BB by comparing transformed samples
Hope that helps, Kamal
Please sign in to leave a comment.
Comments
1 comment