The value indicating the difference between prediction of the values and the correct value.

  • Low bias: causes Underfitting in the model.
  • High bias: causes error and inaccuracy in training and testing of model.

Bias can be seen as the error introduced by approximating the true underlying function.


Variance of the model is the variability of it’s prediction and indicates spread of data.

  • Low variance: Is often considered noise and irrelevant to the prediction.
  • High variance: causes Overfitting in the model.

    Variance can be seen as a model's sensitivity to changes in the training dataset.


Bias-Variance Trade off

It's the relationship between the expected test error and Bias and Variance(Statistics). The zone between “high bias and low variance” model which is too simple, and “high variance and low bias” model which is too complex is call bias-variance trade off and offers best performance on both training data and testing data.
ℹ Tradeoff generally means increasing one parameter would lead to decreasing of other.