Databricks Machine Learning Associate Question 599
Select 3You are training a binary classification model to predict whether a customer will make a purchase (1) or not (0). After evaluating the model, you observe the following metrics: F1-score = 0.85, Log Loss = 0.34, AUC-ROC = 0.92. Based on these metrics, which of the following statements are true about the model's performance?
- A
The model has a good balance between precision and recall.
- B
The model's predictions are well-calibrated with low uncertainty.
- C
The model performs very poorly at distinguishing between the two classes.
- D
The model has a high ability to distinguish between positive and negative classes.
- E
The Log Loss value indicates that the model is overfitting.
Show answer and explanation
Correct answers: A, B, D
Explanation
The metrics provided (F1-score, Log Loss, and AUC-ROC) collectively indicate that the model has a good balance between precision and recall, well-calibrated probabilistic predictions, and excellent ability to distinguish between the two classes. There is no evidence of overfitting or poor performance based on the given metrics.
- A. Correct.
The F1-score of 0.85 suggests a good balance between precision and recall, as F1 is the harmonic mean of these two metrics.
- B. Correct.
Log Loss of 0.34 indicates that the predicted probabilities are well-calibrated and have low uncertainty. Lower Log Loss values represent better-calibrated models.
- C. Incorrect.
The AUC-ROC of 0.92 indicates that the model performs well at distinguishing between the two classes, so this statement is incorrect.
- D. Correct.
AUC-ROC of 0.92 confirms that the model has a high ability to distinguish between positive and negative classes, as an AUC-ROC closer to 1 is ideal.
- E. Incorrect.
The Log Loss value of 0.34 does not suggest overfitting. Overfitting would typically be indicated by a large gap between training and validation performance, which is not mentioned here.