look I have this gigantic dataset and I want to check if there is any relationship between a numerical feature and my categorical target (2 groups). My numeric feature does not follow the normal distribution and I shoukld use a Non-Prametric test to evaluate this condidtion. When I was doing my research I found the Test Mann Whitney U that fits very well on my situation. The problem is, I have also read that those statistical tests are very sensitive when you a use them on a dataset composed of a lot of records, so what should I do? Should I use the test on all of my data or should I take a sample first?
I tried to use all my data but the results seem to allways reject the null Hypothesis...
