Underfitting :
A model with too few parameters (e.g., a linear model for a nonlinear relationship) will have high bias (poor fit) and low variance (predicts the same even for different datasets).
- High Bias
- Low Variance
Overfitting :
A model with too many parameters (e.g., high-degree polynomial) will have low bias (very good fit on training data) and high variance (fails to generalize well to test data).
- Low Bias
- High Variance