Which two metrics are most appropriate for comparing the accuracy of different machine learning models predicting house prices?

When comparing the accuracy of different machine learning models predicting house prices, focusing on metrics that specifically assess regression models is crucial since house price prediction is a regression task rather than a classification task.

Mean Absolute Error (MAE) is a relevant choice because it measures the average magnitude of the errors in a set of predictions, without considering their direction. This metric indicates how close the predictions are to the actual house prices, providing a straightforward interpretation of model performance. Lower MAE values indicate better-performing models, making it an effective choice for this context.

R-squared (Coefficient of Determination) is another appropriate metric for this scenario. It indicates the proportion of the variance in the dependent variable (house prices) that can be explained by the independent variables in the model. A higher R-squared value signals a better fit of the model to the data, making it a valuable metric when evaluating the performance of different models in predicting house prices.

In contrast, Cross-Entropy Loss is primarily used in classification tasks to quantify the difference between the predicted probabilities of classes and the true class labels. The F1 Score is also dedicated to classification concerning the balance between precision and recall. Therefore, these metrics are not suitable for evaluating regression models like those predicting house prices.