Multiple Linear Regression for Reconstruction of Gene Regulatory Networks in Solving Cascade Error Problems

Table 1

Decision tree.

Nature of data								Chosen model type	Recommended model
	Independent variables (predictors)		Dependent variables (response variables)			Model
	Continuous	Categorical	Continuous	Restricted	Multivariable	Linear	Nonlinear
()	(OR)	(OR)						Fitted model coefficients	Linear regression
()	(OR)	(OR)						Fitted model and fitted coefficients	Stepwise regression
()	(OR)	(OR)				(generalized)		Fitted generalized linear model coefficients	Generalized linear models
()			(nonlinear)					Fitted nonlinear model coefficients	Nonlinear regression
()								Ridge/LASSO/elastic net regression	Ridge/LASSO/elastic; net regression
()	(correlated)							Fitted model and fitted coefficients	Partial Least Squares
()	(OR)	(OR)						Nonparametric model	Classification trees, Regression trees, Ensemble methods
()								ANOVA	ANOVA
()								Fitted multivariate regression model coefficients
()								Fitted mixed-effects model coefficients	Mixed-effects model

Note: cells with “” indicate the type of variables that suit the nature of GRN. The recommended models are marked with asterisks ().