Скачать или смотреть Non-linear Analysis Part-1 Introduction

This first video establishes a crucial three-step workflow.
1.Run a standard linear regression model to set a baseline.
2. Rigorously test the validity of the linear assumption by performing Ramsey's RESET test. This formal check determines if significant non-linearity is present in the data.
3. If the test suggests non-linearity, the third step guides to the advanced modeling techniques:
MARS (Multivariate Adaptive Regression Splines) and
GAM (Generalized Additive Models), providing flexible alternatives to capture those complex, non-linear relationships that a simple linear model misses

Base Paper :Multivariate Adaptive Regression Splines" by Jerome H. Friedman
https://projecteuclid.org/journals/an...