A company built an image classifier to detect plant diseases from leaf photos and wants to know the proportion of images the model labeled correctly. Which evaluation metric should they use?

Accuracy is the correct metric because it directly measures the proportion of correctly classified instances (both true positives and true negatives) out of the total number of instances. For an image classifier determining plant diseases, the company wants to know how often the model is right, which is precisely what accuracy provides. R-squared score is used for regression tasks to assess how well a model's predictions approximate real-world values. Root Mean Squared Error (RMSE) is also a regression metric that quantifies the average magnitude of the errors. Learning rate is a hyperparameter used in training machine learning models, not an evaluation metric.