evaluate_by_loss {predbayescor} | R Documentation |
calculating the total loss of prediction results
Description
Calculates the average loss of predictions based on threshold with threshold
. Note that this threshold has 1-1 mapping with the ratio of the loss of assigning 0 to 1 to the loss of assigning 1 to 0: threshold=1-1/(1+ratio).
Usage
evaluate_by_loss ( y.true, pred.prob, threshold=0.5)
Arguments
y.true |
a vector containing the true response. |
pred.prob |
a vector containing the predictive probabilities. |
threshold |
When predictive probability is greater than threshold , the response is predicted as 1. |
Value
loss |
the average loss, with attrib "sd" storing the estimate of the standard error of this loss. |
[Package
predbayescor version 1.1-4
Index]