Hi. Excellent video Dr.Kamper. I was wondering could you do a video on MARS (Multivariate adaptive regression spline) algorithm? Also as you said estimating linearity from visualization and fit could be subjective, so what would be the best way to prove (objectively) that my data is linear or non-linear. I usually apply multilinear regression & polynomial regression and see which drives my rsquare(r2) up. I am just wondering if there is any other way to deal with this? Thank you! Huge fan of your work by the way!
Hey Phani! Very happy you like my work! I wish I had more time to make videos (like the MARS one you ask about). To answer your question: it is very difficult to conclusively prove that the data is definitely linear or non-linear. But one way to do this is to measure the squared loss on some held-out data. Definitely have a look at my other video on evaluating models: ruclips.net/video/aXRDdjK-hI4/видео.html.
I tried finding out what the RBF meant and what it did, but most tutorials failed me. Your vid was clearly gold. Thanks for your effort.
Big pleasure!! Very happy it helped! :)
Thank you saved my soul
Hi. Excellent video Dr.Kamper. I was wondering could you do a video on MARS (Multivariate adaptive regression spline) algorithm? Also as you said estimating linearity from visualization and fit could be subjective, so what would be the best way to prove (objectively) that my data is linear or non-linear. I usually apply multilinear regression & polynomial regression and see which drives my rsquare(r2) up. I am just wondering if there is any other way to deal with this? Thank you! Huge fan of your work by the way!
Hey Phani! Very happy you like my work! I wish I had more time to make videos (like the MARS one you ask about). To answer your question: it is very difficult to conclusively prove that the data is definitely linear or non-linear. But one way to do this is to measure the squared loss on some held-out data. Definitely have a look at my other video on evaluating models: ruclips.net/video/aXRDdjK-hI4/видео.html.
Do you have a regression for a function Gaussian and polynomial together?