conversion_expected_loss: Find the expected loss in a conversion experiment

In dgrtwo/splittestr: Bayesian and frequentist A/B split testing

Find the expected loss in a conversion experiment

Description

This is a wrapper around expected_loss. It applies a prior and calculates in terms of successes and totals of A and B, instead of the beta distribution. Note that the "error" is treatment A being better than B.

Usage

conversion_expected_loss(
  sA,
  nA,
  sB,
  nB,
  prior_alpha = 1,
  prior_beta = 1,
  approx = FALSE,
  ...
)

Arguments

prior_alpha	prior parameter to use for alpha (default uniform)
prior_beta	prior parameter to use for beta (default uniform)
approx	whether to use a normal approximation for h
...	Extra arguments, passed on to expected_loss
success_A	number of successes (e.g. clicks) in condition A
total_A	total number of exposures in A
success_B	number of successes (e.g. clicks) in condition B
total_B	total number of exposures in B

dgrtwo/splittestr documentation built on May 15, 2022, 3:16 p.m.