Deeply digging the interaction effect in multiple linear regressions using a fractional-power interaction term.

ABSTRACT: In multiple regression Y ~ β₀ + β₁X₁ + β₂X₂ + β₃X₁ X₂ + ɛ., the interaction term is quantified as the product of X₁ and X₂. We developed fractional-power interaction regression (FPIR), using βX₁ ^M X₂ ^N as the interaction term. The rationale of FPIR is that the slopes of Y-X₁ regression along the X₂ gradient are modeled using the nonlinear function (Slope = β₁ + β₃MX₁ ^M-1 X₂ ^N), instead of the linear function (Slope = β₁ + β₃X₂) that regular regressions normally implement. The ranges of M and N are from -56

SUBMITTER: Li X 

PROVIDER: S-EPMC7549115 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature