In kernelized linear regression, we transform the covariate coordinates (x) such that the predictor (y) becomes a linear function of the covariates. I wonder if I can find such a transformation that does this for 1/x (which is discontinuous) or any other discontinuous function.
Asked
Active
Viewed 20 times
0
-
Welcome to MSE. Please read this text about how to ask a good question. – José Carlos Santos Jun 18 '21 at 09:52
-
Do you have good reasons to fit across this singularity ? – Jun 18 '21 at 10:10
-
The answer is in the question: set $x'=\dfrac1x$. – Jun 18 '21 at 10:17
-
Dear Yves, I would normally do that but my friend is convinced. I can not show him what such a coordinate system would look like. (Neither of us is math major). – thisisme Jun 18 '21 at 13:20