Databricks Machine Learning Associate Question 168
Single answerYou are tasked with building a binary classification model using Databricks. The model predicts whether a customer will churn (1) or not (0). After evaluating the model, you decide to adjust the decision threshold from the default 0.5 to 0.7. Which of the following is the most likely outcome of this change?
- A
The number of false positives will increase.
- B
The number of false negatives will increase.
- C
The overall accuracy of the model will improve regardless of the dataset.
- D
The precision of the model will likely improve, but recall may decrease.
Show answer and explanation
Correct answer: D
Explanation
When the decision threshold is increased from 0.5 to 0.7, the model becomes more conservative in predicting the positive class (customer churn). This typically results in fewer false positives, leading to an improvement in precision. However, it also increases the likelihood of false negatives, reducing recall. The overall accuracy may or may not change depending on the dataset, but precision and recall are the most directly affected metrics.
- A. Incorrect.
Reducing the threshold (not increasing it) would lead to more predictions of the positive class, potentially increasing false positives. Increasing the threshold decreases false positives.
- B. Incorrect.
Increasing the threshold makes the model stricter in predicting the positive class, leading to more false negatives (missed positive predictions). However, this is not the most comprehensive answer to the question.
- C. Incorrect.
The overall accuracy of the model depends on the dataset's class distribution and the balance between true positives, true negatives, false positives, and false negatives. Changing the threshold does not guarantee an improvement in accuracy.
- D. Correct.
Adjusting the threshold to 0.7 makes the model more selective in predicting the positive class, which can improve precision (fewer false positives) but may reduce recall (more false negatives). This is the most likely outcome of the change.