I am trying to down sample a data set and keep the same frequency distribution as one of the column. The approach is 1) determine the baseline frequency distribution, 2) use the baseline frequency distribution to sample the row range, 3) use the sampled row range to select rows from the baseline data frame, 4) compare the baseline and down sampled frequency distributions. Here are two examples. In both examples, the events with the highest probability are oversampled, and the remaining events are under-sampled. I am trying to down sample a data set and keep