Description
A/B testing is a tool that allows to check whether a certain causal relationship holds. For example, a data scientist working for an e-commerce platform might want to increase the revenue by…
Summary
- A/B testing is a tool that allows to check whether a certain causal relationship holds.
- Maybe, if we had run the experiment an other day, only l persons instead of k would have clicked on the link and then our estimate would have been In fact, to take into account all possible cases of the binomial count, we define the maximum likelihood estimator of p, as the random variable where Y has a binomial B(n, p) distribution.
- So, we replace the observed value k by the random variable Y that gave rise to this observation to be able to take into account all possible outcomes of the experiment.
- This allows us to think more generally about our experiment.