X% chance to not generate row at all (based on other dataset)

I have a Dataset I generated (Table A) with N records. For each record in the Table A, I want to create 1-4 records in the new table (Table B). This can be generated differently every time, that does not matter. I just do not want 4 Table B records for every Table A record every time. Some Table A records will have 4 Table B records, but not every single one. Is this possible? I dont want N*4 records in table B every time, and I need the table to be varied, with some Table A records having more entries in Table B than others.

I currently just set the chance of every column in Table B to be 30%, then told my IDE to ignore all errors so it skips over the error of the fields being null when they should not be. This does do what I want, but creates other problems, like downloaded CSV having blank records as well, which in turn breaks another dataset I want to generate based on that. It would be best to get this worked out so that each record in Table B has a X% chance of not being created at all.

I have been think about this some more. I think two useful features would be 1) a Skip Row ruleset. This could be a combination of a percentage chance, and even custom rules using the normal syntax. This could also be combined with a min row count to work with the next feature. That is 2) an AUTO amount of rows generated instead of a fixed amount, which would be allowed if at least one column was iterable. So, if I want to create one record in table B for each record in table A, I set the foreign key column to sequential, it creates that many rows and I don’t have to set it explicitly. I can also give each row a chance to skip, and set a foreign key column referencing table A with N records to Cartesian and another custom list column with M list items to Cartesian and set the number of rows to AUTO, and I’ll automatically get between min-MxN rows generated for me, completely dynamic and completely based on table A.