regularization_plot: Plots coefficients from regularizing various models.

In UBC-MDS/ezmodelR:

Description

Plots coefficients from regularizing various models.

Usage

regularization_plot(model, lambda, tol = 1e-07, x, y)

Arguments

model:	String specifying lasso, ridge regression, or logistic regression with L2-regularization. Argument should be one of "lasso", "ridge", or "logistic".
lambda:	Vector of penalty constant(s) multiplying the regularization term. Larger value corresponds to stronger regularization.
tol:	Coefficients less than this will be treated as zero.
x:	n x d dataframe of features.
y:	n x 1 dataframe of response values.

Value

ggplot object. Plot returned depends on length of lambda. length(lambda)==1: Plot displays magnitude of model coefficients, where coefficients with magnitude less than 'tol' are treated as zero. length(lambda)>1: Plot displays counts of nonzero coefficients in each model, where coefficients with magnitude less than 'tol' are treated as zero.

Examples

X <- mtcars[-1]
Y <- data.frame(mtcars$mpg)

regularization_plot('ridge', lambda=2^c(1,0,1), x=X, y=Y)
regularization_plot('lasso', lambda=2, x=X, y=Y)

UBC-MDS/ezmodelR documentation built on May 25, 2019, 1:35 p.m.