Scoring: Evaluate predictions

In multiridge: Fast Cross-Validation for Multi-Penalty Ridge Regression

Evaluate predictions

Description

Evaluates predictions by a score suitable for the corresponding response

Usage

Scoring(lp, Y, model = NULL, score = ifelse(model == "linear", "mse", "loglik"),
  print = TRUE)

Arguments

lp	Numerical vector. Linear predictor.
Y	Response vector: numeric, binary, factor or survival.
score	Character. See Details.
model	Character. Any of c("linear", "logistic", "cox"). Is inferred from Y when NULL.
print	Boolean. Should the score be printed on screen.

Details

Several scores are allowed, depending on the type of output. For model = "linear", score equals any of c("loglik","mse","abserror","cor","kendall","spearman"), denoting CV-ed log-likelihood, mean-squared error, mean absolute error, Pearson (Kendall, Spearman) correlation with response. For model = "logistic", score equals any of c("loglik","auc", "brier"), denoting CV-ed log-likelihood, area-under-the-ROC-curve, and brier score a.k.a. MSE. For model = "cox", score equals any of c("loglik","cindex"), denoting CV-ed log-likelihood, and c-index.

Value

Numerical value.

See Also

CVscore for obtaining the cross-validated score (for given penalties), and doubleCV to obtain doubly cross-validated linear predictors to which Scoring can be applied to estimated predictive performance by double cross-validation. A full demo and data are available from:
https://drive.google.com/open?id=1NUfeOtN8-KZ8A2HZzveG506nBwgW64e4

multiridge documentation built on June 13, 2022, 5:07 p.m.