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

evaluation metrics

Innocent Iguana answered on February 10, 2023 Popularity 10/10 Helpfulness 3/10

answer evaluation metrics

evaluation metrics

Comment

Way of measurement for specific models that tells how good/bad a model's
performance is. It is the process of understanding how good or bad a model is. 
Generally insample data (Training data) is used to build the model and evaluation
is done on out-of-sample data (Testing data) to see the performance.

For classification, accuracy is not the best measurement of performance.
WHY? let's say a spam email detection model was developed based on a dataset 
that has 90% right email and 10% spam email. Even if we blindly tell that all 
emails (100%) are not spam then our model cannot catch a single spam email but 
still gives a 90% accuracy. Because 90% not-spam was correctly detected while 
10% was incorrect. However, the whole purpose behind building the model was to 
catch some spams from those 10%.

Some classification evaluation metrics are:
1. Jaccard Index
2. Log loss
3. Sensitivity
4. Specificity
5. F1-Score (The sweet spot between sensitivity and specificity) 

For regression, model evaluation metrics determine how much of error (residuals)
does the model produce when determining a continuous value for unseen data.
The popular evaluation metrics for regression model are:
1. MSE
2. R-squared

Popularity 10/10 Helpfulness 3/10 Language whatever

Source: Grepper

Tags: evaluation metrics whatever

Link to this answer
Share Copy Link

Contributed on Feb 10 2023

Innocent Iguana

0 Answers Avg Quality 2/10

evaluation metrics

Contents

More Related Answers

evaluation metrics

Grepper

Documentation

Social

Legal

Contact

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