Databricks Machine Learning Professional Question 220
Single answerYou are deploying a machine learning model for predicting customer churn, and after monitoring the model for several months, you notice that the distribution of input features has shifted significantly compared to the training data. However, the relationship between the features and the target variable remains consistent. Which type of drift is most likely occurring?
- A
Concept drift
- B
Covariate drift
- C
Label drift
- D
Model degradation
Show answer and explanation
Correct answer: B
Explanation
The scenario describes a situation where the input features' distribution has shifted, but the relationship between the features and the target variable has remained stable. This is characteristic of covariate drift, where the issue lies in the input data rather than the model's learned relationships.
- A. Incorrect.
Concept drift occurs when the relationship between input features and the target variable changes over time, which is not the case here since the relationship remains consistent.
- B. Correct.
Covariate drift refers to changes in the distribution of input features while the relationship between the features and the target variable remains consistent. This matches the scenario described.
- C. Incorrect.
Label drift occurs when the distribution of the target variable changes over time, but there is no indication of this in the scenario provided.
- D. Incorrect.
Model degradation typically refers to a decline in model performance over time, often due to drift, but it is not a specific type of drift.