Most Common Loss function and Evaluation metrics in machine learning(Regression)

It’s been a long, I have not shared any knowledge content.

Now I plan to share some useful content for Machine learning /Data Scientist engineer on timely basis.

Content will mostly include:

1-   Machine Learning

2-   Deep Learning

3-   Natural Language Processing

4-   Artificial Intelligence

5-   Latest Research / Research Papers

Today, I will start with the most fundamental Evaluation metrics for regression.

MAE: Mean Absolute Error calculates the average difference between the calculated values and actual values.

Less bias for large error (outliers) but also May not be adequately representing large errors. It does put large weight on outlier because error weighted on linear scale.

MSE: It is simply the average of the square of the difference between the original values and the predicted values.

Highly biased for large errors. It’s a good loss function if we can take care of outliers. It put large weight on outliers.

RMSE: Same as MSE as this is just a square root of MSE.

Tips: Use MSE if outlier important.