Databricks Machine Learning Professional Question 234
Single answerA company deploys a machine learning model to predict customer churn based on customer behavior data. After six months, the model's predictions become less accurate, and the business team observes an increase in false negatives for churn predictions. Which of the following best explains the likely cause of this issue?
- A
The model's hyperparameters were not tuned adequately prior to deployment.
- B
The training data used to build the model was too small to capture all potential patterns.
- C
A concept drift occurred, changing the relationship between customer behavior features and churn outcomes.
- D
The model was not optimized for the deployment environment's computational limitations.
Show answer and explanation
Correct answer: C
Explanation
Concept drift occurs when the statistical properties of the input data or the relationship between features and the target variable change over time. This can lead to a decline in model performance, as the model is no longer aligned with the current data distribution. In this scenario, the customer behavior patterns likely evolved, causing the model's predictions to become less accurate.
- A. Incorrect.
While hyperparameter tuning can improve model performance, it does not directly explain why the model's accuracy deteriorates over time after deployment.
- B. Incorrect.
Using a small training dataset can lead to poor initial performance, but it does not explain why the model's performance degrades over time after deployment.
- C. Correct.
Concept drift refers to changes in the statistical properties of input data or their relationship to the target variable over time. This is the most likely reason for the observed decline in model accuracy and the increase in false negatives.
- D. Incorrect.
Deployment environment computational limitations may cause latency or resource issues, but they do not explain why the model's accuracy declines over time.