Re: Nonparametric regression: Like parametric regression, but not

Hi everyone
Happy I'm here. I was seeking the non parametric equivalence of my 48 observations time motion study with a slightly negatively skewed time distribution
I see a kernel regretion like this

npregress kernel hectoliters taxlevel, vce(bootstrap, reps(100) seed(123))

Could I please know the meaning of replications and seed and their implications on results. What is the advantage of 100 replications over 50? What is the best practice?

Thanks