loocv: Leave-one-out cross validation

In ddgraph: Distinguish direct and indirect interactions with Graphical Modelling

Description

Leave-one-out cross validation systematically leaves out one row from the data, retrains the classifier and then uses the retrained classifier to make a prediction for the left-out row.

Usage

loocv(data, train.fun, eval.fun, verbose=FALSE)

Arguments

data	The data.frame with data. Columns are variables, rows are observations.
train.fun	The training function that takes the data without one of the rows left out.
eval.fun	The prediction function that takes the trained model and the left out data point.
verbose	If to print progress indication

Value

A vector of length nrow(data) containing predictions from eval.fun when each row is left out once

ddgraph documentation built on Nov. 17, 2017, 10:50 a.m.