I'm using SQL in Python.

I'm running a sample set of 100 from a population of data (1mn+ rows). I am using the following function:

RAND() <= 0.0025 LIMIT 100

I am unable to get a good sample as the fields are duplicating (there is no duplication in data). If i change the seed number- it gets better.

I tried searching what to put as seed number & how seed number works but couldn't find anything relevant.

I want to know why setting 0.0025 would be better than 0.45 in an instance?