residual sum of squares

$\text{RSS}={\sum }_{i=1}^n(y_i-{\hat{y}}_i{)}^2.$ where

• $y_i$ is the observed value
• ${\hat{y}}_i$ is the predicted value

Note: Squaring the error will more heavily penalize a few large errors than many small errors. Even if the sum of errors is the same.