Regression Evaluation Metrics
Regression Evaluation Metrics are used to measure:
How well a regression model performs
These metrics compare:
-
Actual values
-
Predicted values
and calculate:
Prediction error
Why Evaluation Metrics are Important
After training a regression model:
We must check how accurate the predictions are
Evaluation metrics help us:
-
Measure model performance
-
Compare models
-
Detect errors
-
Improve prediction quality
Example
Suppose:
| Actual Price | Predicted Price |
|---|---|
| 100 | 90 |
| 200 | 210 |
| 300 | 280 |
The model predictions are close, but not perfect.
Evaluation metrics calculate:
How much error exists
Common Regression Evaluation Metrics
| Metric | Full Form |
|---|---|
| MAE | Mean Absolute Error |
| MSE | Mean Squared Error |
| RMSE | Root Mean Squared Error |
| R² Score | Coefficient of Determination |
1. Mean Absolute Error (MAE)
MAE calculates:
Average absolute error
Formula:
MAE = mean(|Actual - Predicted|)
MAE Example
| Actual | Predicted | Error |
|---|---|---|
| 100 | 90 | 10 |
| 200 | 210 | 10 |
| 300 | 280 | 20 |
Calculate MAE
(10 + 10 + 20) / 3
40 / 3
13.33
Final MAE
MAE = 13.33
Lower MAE:
Better model
2. Mean Squared Error (MSE)
MSE calculates:
Average squared error
Formula:
MSE = mean((Actual - Predicted)²)
MSE Example
| Actual | Predicted | Error | Error² |
|---|---|---|---|
| 100 | 90 | 10 | 100 |
| 200 | 210 | -10 | 100 |
| 300 | 280 | 20 | 400 |
Calculate MSE
(100 + 100 + 400) / 3
600 / 3
200
Final MSE
MSE = 200
Large errors get:
Higher punishment
because errors are squared.
3. Root Mean Squared Error (RMSE)
RMSE is:
Square root of MSE
Formula:
RMSE = √MSE
RMSE Calculation
Suppose:
MSE = 200
Then:
RMSE = √200
RMSE ≈ 14.14
Why RMSE is Useful
RMSE converts:
Squared error back to original units
This makes interpretation easier.
4. R² Score (Coefficient of Determination)
R² Score measures:
How well the model explains the data
Range:
| R² Score | Meaning |
|---|---|
| 1 | Perfect prediction |
| 0 | No learning |
| Negative | Very poor model |
R² Formula
R² = 1 - (SSR / SST)
Where:
-
SSR → Sum of Squared Residuals
-
SST → Total Sum of Squares
Understanding R²
Suppose:
R² = 0.90
This means:
90% of the variation is explained by the model
Which Metric is Better?
| Metric | Best Use |
|---|---|
| MAE | Simple average error |
| MSE | Penalizes large errors |
| RMSE | Error in original units |
| R² | Overall model performance |
Important Understanding
Lower values are better for:
-
MAE
-
MSE
-
RMSE
Higher values are better for:
-
R² Score
Practical Example Using Python
Step 1: Import Libraries
from sklearn.metrics import (
mean_absolute_error,
mean_squared_error,
r2_score
)
import numpy as np
Step 2: Actual and Predicted Values
actual = np.array([100, 200, 300])
predicted = np.array([90, 210, 280])
Step 3: Calculate MAE
mae = mean_absolute_error(actual, predicted)
print("MAE:", mae)
Step 4: Calculate MSE
mse = mean_squared_error(actual, predicted)
print("MSE:", mse)
Step 5: Calculate RMSE
rmse = np.sqrt(mse)
print("RMSE:", rmse)
Step 6: Calculate R² Score
r2 = r2_score(actual, predicted)
print("R² Score:", r2)
Example Output
MAE: 13.33
MSE: 200
RMSE: 14.14
R² Score: 0.97
Advantages of Evaluation Metrics
-
Measure prediction quality
-
Compare multiple models
-
Detect poor models
-
Improve model selection
Real-World Applications
| Industry | Usage |
|---|---|
| Finance | Stock prediction accuracy |
| Healthcare | Disease prediction evaluation |
| Real Estate | House price prediction quality |
| Sales | Revenue forecasting accuracy |
Important Points
1. MAE measures average absolute error.
2. MSE squares errors and penalizes large mistakes.
3. RMSE converts squared error into original units.
4. R² Score measures goodness of fit.
5. Lower error metrics indicate better performance.
Summary
Regression Evaluation Metrics help measure the accuracy and quality of regression models. Metrics such as MAE, MSE, RMSE, and R² Score compare predicted values with actual values and help determine how well the model performs on unseen data.
Keywords
Regression Evaluation Metrics, MAE, MSE, RMSE, R2 Score, Regression Accuracy, Mean Absolute Error, Mean Squared Error, Root Mean Squared Error, Coefficient of Determination, Model Evaluation, Regression Performance Metrics, Machine Learning Metrics, Regression Error Calculation, Evaluation Metrics in Machine Learning
