I have a dataset set called iso3166CountryCodes with 242 records and I wanted it to deliver a high proportion of one country compared to others. I set the column to numeric isocode and then went to edit the customer distribution. I found that the 826 records appeared many times. When I changed its weight rule to 100, all of the repeated 826s changed to 100 too.
I have since experimented by removing the frequency column and all the repeated 826s disappeared - I assume that records are inserted into the dataset in the proportion of the frequency column.