R functions that let you measure if a A/B test :
- has the same conversion rate
- that a version is more efficiente than another one.
It use the N - 1 two-proportion test algorithm.
Check http://www.measuringusability.com/ab-calc.php for a quick test.
You have 2 functions :
- same_conv : Probability that the two versions have the same conversion rate as (in %),
- conv_interval : Return the lowest & highest conversion rates based on a confidence interval set (in %). It returns a list : $low & $high.
arg | Desc | Type | Mandatory |
---|---|---|---|
visits A | Total number of visits of version A | Num | Yes |
visits B | Total number of visits of version B | Num | Yes |
conversions A | Total number of conversions of version A | Num | Yes |
conversions B | Total number of conversions of version B | Num | Yes |
arg | Desc | Type | Mandatory |
---|---|---|---|
confidence interval | Confidence interval (expressed as %. Ex : 0.90) | Num | Yes |
visits A | Total number of visits of version A | Num | Yes |
visits B | Total number of visits of version B | Num | Yes |
conversions A | Total number of conversions of version A | Num | Yes |
conversions B | Total number of conversions of version B | Num | Yes |
- conv_interval(args*)$low
- conv_interval(args*)$high