Metric	Full Form	What It Measures	Units	Sensitive to Outliers	When to Use	When NOT to Use
MAE	Mean Absolute Error	Average absolute difference between actual and predicted values	Same as target (₹, days, marks)	Low	When all errors matter equally; data has outliers; want easy interpretation	When large errors are very costly
MSE	Mean Squared Error	Average of squared prediction errors	Squared units (₹², days²)	Very High	During model training and optimization; mathematical convenience	For interpretation or reporting to humans
RMSE	Root Mean Squared Error	Square root of MSE (penalizes large errors)	Same as target	High	When large errors are dangerous (medical, finance, engineering)	When outliers dominate and typical error matters
MAPE	Mean Absolute Percentage Error	Average error as a percentage of actual value	Percentage (%)	Medium	When comparing across different scales; business & forecasting	When actual values can be zero or very small

One-Line Intuition for Each Metric

• MAE → “On average, how much am I wrong?”
• MSE → “How badly do big mistakes hurt my model?”
• RMSE → “How serious are large errors, in real units?”
• MAPE → “How wrong am I in percentage terms?”

Scenario	Best Metric
Simple interpretation	MAE
Safety-critical systems	RMSE
Model training (optimization)	MSE
Business / sales forecasting	MAPE
Data has outliers	MAE + RMSE together
Comparing models fairly	RMSE + MAPE

MAE measures average absolute error and is easy to interpret. MSE squares errors and is mainly used during model training. RMSE is the square root of MSE and penalizes large errors strongly, making it suitable for critical applications. MAPE expresses error in percentage terms and is useful for comparing predictions across different scales but fails when actual values are zero.