selection.metrics: Selection metrics

In gesso: Hierarchical GxE Interactions in a Regularized Regression Model

Description

Calculates principal selection metrics for the binary zero/non-zero classification problem (sensitivity, specificity, precision, auc).

Usage

selection.metrics(true_b_g, true_b_gxe, estimated_b_g, estimated_b_gxe)

Arguments

true_b_g	vector of true main effect coefficients
true_b_gxe	vector of true interaction coefficients
estimated_b_g	vector of estimated main effect coefficients
estimated_b_gxe	vector of estimated interaction coefficients

Value

A list of principal selection metrics

b_g_non_zero	number of non-zero main effects
b_gxe_non_zero	number of non-zero interactions
mse_b_g	mean squared error for estimation of main effects effect sizes
mse_b_gxe	mean squared error for estimation of interactions effect sizes
sensitivity_g	recall of the non-zero main effects
specificity_g	recall of the zero main effects
precision_g	precision with respect to non-zero main effects
sensitivity_gxe	recall of the non-zero interactions
specificity_gxe	recall of the zero interactions
precision_gxe	precision with respect to non-zero interactions
auc_g	area under the curve for zero/non-zero binary classification problem for main effects
auc_gxe	area under the curve for zero/non-zero binary classification problem for interactions

Examples

data = data.gen()
model = gesso.cv(data$G_train, data$E_train, data$Y_train)
gxe_coefficients = gesso.coef(model$fit, model$lambda_min)$beta_gxe                
g_coefficients = gesso.coef(model$fit, model$lambda_min)$beta_g  
selection.metrics(data$Beta_G, data$Beta_GxE, g_coefficients, gxe_coefficients)

gesso documentation built on Nov. 30, 2021, 9:09 a.m.