Here we are trying to predict the value of *one observation*.

The red area is the area where the prediction *should* lie, that is, the true value of the observation lies roughly in the red region.

a) Top left: This model predicts values which lie spot on the red region(low bias) on evaluation with different observations (the number of blue points represents that) and the predictions are close to each other(low variance).

b) Top right: This model predicts values which lie around the red region (low bias) on evaluation with different observations(the number of blue points represents that) and the predictions are far away from each other(high variance).

c) Bottom left: This model predicts values which lie around the blue region (high bias, that is this model roughly predicts in area which is far from the true area) on evaluation with different observations (the number of blue points represents that) and the predictions are far away from each other (high variance).

d) Bottom right: This model predicts values which lie around the white region(high bias, that is this model roughly predicts in area which is far from the true area) on evaluation with different observations(the number of blue points represents that) and the predictions are far away from each other(high variance).

Regards