Mean Absolute Percentage Error (MAPE)

INFO

Measures the percentage error between actual and predicted values, useful for interpretability.

How It Works

$$\text{MAPE}=\frac{100}{n}\sum _{i=1}^n\left|\frac{y_i-{\hat{y}}_i}{y_i}\right|$$

• Expresses error as a percentage
• Undefined when $y_i=0$

What to Look For

• Lower MAPE = better performance
• Easy to interpret across domains
• Avoid when actual values can be zero

Application Models

• Logistic Regression
• Time Series Analysis