Let's create a representative sample of a dataset for a given predictive analytics problem, that's random enough but always the same, every time you run the sampling algorithm