The beta distribution is used primarily to model recovery rates. The beta distribution can be skewed or symmetric. The beta distribution depends upon two inputs: mean and variance. The Kernel distribution is preferred to the beta estimation for bimodal samples (bimodal means there are two modes).