Mean Squared Error (MSE)

INFO

Measures the average squared difference between actual and predicted values, used for regression models.

How It Works

$$\text{MSE}=\frac{1}{n}\sum _{i=1}^n{\left(y_i-{\hat{y}}_i\right)}^2$$

• Penalizes larger errors more than Mean Absolute Error (MAE)
• Sensitive to outliers

What to Look For

• Lower MSE = better performance
• Use when large errors are especially undesirable
• Not directly interpretable in target units

Application Models

• Deep Neural Network (DNN)
• Recurrent Neural Network (RNN)