Linear Regression

No.11257863 ViewReplyOriginalReport
I have some shit and some questions. I never took stats and I am retarded.

let Kn be a set of pairs where each pair is k1 and k2
to find all pairs: Kn/2(Kn+1)
at each Kn for every 2t: f((t1+t2)/2) = Mn
where Mn is a set of the mean of every two consecutive t's
now if n points from Mn is of distance d from ax+b then a linear relationship can be evaluated between a k1 and k2 within Kn

How many data points should I give a shit about and what is the distance from the regression line that makes the pair linear ?

Also, why cant I just take the absolute difference between two arrays then take the standard deviation of that result then look at the margin of error?