2

I'm using R. My dataset has about 39 different Variables/Vektors and each has about 80 entries.Furthermore I have 1 Variable with 80 entries I want to maximize (lets call it Y). I assume that there are some correlations in my data and I want to find out what values my 39 variables should have in order for my Y to be maximal. I'm also very sure that not all variables have significant impact on my Y.

What i already did is calculate the correlations of the 39 Variables to my Y and by knowing which correlations where very high/very low I can assume what values they should have to maximize Y. But im not 100% confident that this is the best way to approach my problem because you can't draw big conclusions just from correlation. However I do not know what else I should do and where I should look to solve this problem.

Tim
  • 21
  • 1

1 Answers1

1

I would recommend using rank correlations as some relationships may be non-linear and inspect plots of Y with all independent variables as some may be even non-monotonic. And applying regression tree as a preliminary analysis. Then pick the most promising predictors and combine them in a model - a function which can be maximized by solving for predictors.

Germaniawerks
  • 1,027
  • 1
  • 10
  • 15