NCA-AIIO Question 66
Single answerYou are tasked with comparing the performance of two machine learning models designed to predict housing prices. Model A achieves a Mean Squared Error (MSE) of 25.4 and an R-squared (R²) value of 0.92, while Model B achieves a Mean Squared Error (MSE) of 23.1 and an R-squared (R²) value of 0.89. Which of the following statements best describes the comparison of these models?
- A
Model A performs better overall because it has a higher R-squared (R²) value, indicating it explains more variance in the data.
- B
Model B performs better overall because it has a lower Mean Squared Error (MSE), indicating its predictions are closer to the actual values.
- C
Model A performs better overall because both R-squared (R²) and Mean Squared Error (MSE) are equally important, and its higher R-squared outweighs its slightly higher MSE.
- D
Model B performs better overall because Mean Squared Error (MSE) is the most important metric in regression tasks.
Show answer and explanation
Correct answer: A
Explanation
When comparing regression models, it is essential to consider both Mean Squared Error (MSE) and R-squared (R²). In this scenario, Model A has a significantly higher R-squared (R²), indicating it explains more variance in the data, which is crucial for tasks requiring interpretability. The smaller difference in MSE does not outweigh this advantage, so Model A is the better model overall.
- A. Correct.
Correct. Model A’s higher R-squared (R²) value indicates that it explains more of the variance in the target variable than Model B, making it the better model overall in this scenario.
- B. Incorrect.
Incorrect. While Model B has a lower Mean Squared Error (MSE), the difference in MSE is small and does not outweigh the importance of Model A’s significantly higher R-squared (R²) value.
- C. Incorrect.
Incorrect. This statement misinterprets the trade-off between R-squared (R²) and MSE. R-squared is more relevant for explaining variance in the data, which is critical in this context.
- D. Incorrect.
Incorrect. While MSE is important in regression tasks, it is not always the sole determinant of model performance. In this case, R-squared (R²) provides more insight into the model's explanatory power.