Breaking News: Grepper is joining You.com. Read the official announcement!

log loss

Innocent Iguana answered on February 11, 2023 Popularity 10/10 Helpfulness 4/10

answer log loss

log loss

Comment

It is an evaluation metric that measures the performance of a classification
model based on the probability of output value between 0 and 1.
Basically it is a cost function that estimates the total error of a model
for the training dataset.
It uses kind of normal distribution to judge the likelihood of a result.
It applies penalty when the predicted probability diverges from 
the actual value, making the log-loss value higher for deviated result.
For any given problem, a lower log loss value means better predictions.
In simpler words, it is a probabilistic measure of accuracy.
example: when predicted value is 1 and real value is 1, cost is 0
using "-log(y_pred)" function, when predicted value is 0 and real value is 1,
cost is large.
math: https://miro.medium.com/max/783/1*6oBgYMy4wOls9zGC-frSQg.png

Popularity 10/10 Helpfulness 4/10 Language whatever

Source: Grepper

Tags: loss whatever

Link to this answer
Share Copy Link

Contributed on Feb 11 2023

Innocent Iguana

0 Answers Avg Quality 2/10

Continue with Google

By continuing, I agree that I have read and agree to Greppers's Terms of Service and Privacy Policy.

log loss

Contents

More Related Answers

log loss

Grepper

Documentation

Social

Legal

Contact

Oops, You will need to install Grepper and log-in to perform this action.