Why to choose L1 Loss over MSE?

MSE and L1 both can be used for regression problems. Why to choose L1 Loss over MSE?

1 Like

MSE is Mean squared Error or L2 Loss. It squares the error before taking an average therefore it is becomes very high if our data has outliers.
L1 loss also known as Mean Absolute Error. L1 loss just takes the average of absolute differences of the errors. Therefore it is robust to outliers.
Check this:
https://heartbeat.fritz.ai/5-regression-loss-functions-all-machine-learners-should-know-4fb140e9d4b0

3 Likes