AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Accuracy

  Why it's wrong here

  Accuracy is a classification metric that computes the ratio of correct predictions to total predictions. For regression, however, there are no discrete categories, so 'correct' is not well-defined; continuous predictions rarely match actual values exactly. Using accuracy would essentially always yield zero or a misleadingly low score, providing no useful signal about prediction quality. Regression instead requires error-based metrics like RMSE or MAE that measure how far predictions deviate from actual values.

• ✓

  Root Mean Squared Error (RMSE)

  Why this is correct

  Root Mean Squared Error (RMSE) is the standard evaluation metric for regression models. It is calculated by taking the square root of the average of the squared differences between predicted and actual values, which penalizes large errors more heavily than small ones. A lower RMSE indicates predictions are closer to the true continuous values, and it is expressed in the same units as the target variable, making it directly interpretable. Unlike classification metrics, RMSE naturally handles continuous numeric predictions and is sensitive to outliers, which can be desirable when large errors are especially problematic.

• ✗

  Precision and recall

  Why it's wrong here

  Precision and recall are metrics from binary classification that quantify the model's ability to correctly identify positive instances relative to false positives and false negatives. They require a confusion matrix built from true positives, true negatives, false positives, and false negatives, all of which depend on categorical outcomes. In regression, the output is a continuous variable, so there is no concept of a 'positive' or 'negative' prediction, and no threshold to separate classes. Thus precision and recall cannot be computed or meaningfully interpreted for regression models.

• ✗

  AUC-ROC curve

  Why it's wrong here

  The AUC-ROC curve is designed to evaluate binary classification models by plotting the true positive rate against the false positive rate across various classification thresholds. It relies on predicted probabilities and a binary ground truth, neither of which exists in a regression problem where the target is a continuous numeric value. Regression tasks do not have thresholds or positive/negative classes, so AUC-ROC is entirely inapplicable. Instead, regression evaluation should use error-based metrics such as RMSE, MAE, or R-squared.

Go deeper

Related to this question