tester: Higher-order model testing

In pievos101/DFNET: Network-guided greedy decision forest for feature subset selection

Higher-order model testing

Description

Returns a procedure that can be used as performance metric in train.

Usage

tester(data, target, metric = ModelMetrics::auc, do.predict = TRUE)

Arguments

data	matrix or 3D array. The data to use for testing.
target	numeric vector. The target values to predict.
metric	binary function. A metric to compare actual values with predictions.
do.predict	logical. Should predict be called (TRUE) or $predictions be accessed directly (FALSE)?

Value

A unary function which, takes a ranger-like predictor and evaluates metric with target and the predictions of predictor on data.

pievos101/DFNET documentation built on Dec. 1, 2022, 3:44 p.m.