loss_LL_grad: Loglikelihood Error (gradient function)

In Laurae2/Laurae: Advanced High Performance Data Science Toolbox for R

Description

This function computes the Loglikelihood loss (logloss) gradient per value provided preds and labels probabilities.

Usage

loss_LL_grad(y_pred, y_true)

Arguments

y_pred	The predictions.
y_true	The labels.

Details

Loss Formula : - (y_true * log(y_pred) + (1 - y_true) * log(1 - y_pred))

Gradient Formula : (1 - y_true)/(1 - y_pred) - y_true/y_pred

Hessian Formula : (1 - y_true)/((1 - y_pred) * (1 - y_pred)) + y_true/(y_pred * y_pred)

Value

The gradient of the Loglikelihood Error per value.

Laurae2/Laurae documentation built on May 8, 2019, 7:59 p.m.