Databricks Machine Learning Professional Question 242
Select 3You are monitoring a deployed machine learning model that predicts housing prices. To ensure the model remains reliable, you decide to monitor numeric feature drift using summary statistics. Which of the following statements accurately describes this approach?
- A
Summary statistic monitoring compares distribution properties, like mean and standard deviation, of numeric features between training and production data.
- B
Sudden changes in a numeric feature's summary statistics, such as an increase in the mean or variance, can indicate feature drift.
- C
Summary statistic monitoring is useful for detecting feature drift in categorical features by analyzing frequency distributions.
- D
Tracking summary statistics is computationally efficient compared to more complex drift detection techniques.
- E
Summary statistic monitoring can directly identify the root cause of model performance degradation.
Show answer and explanation
Correct answers: A, B, D
Explanation
Summary statistic monitoring is a simple and efficient method to detect numeric feature drift by analyzing changes in distribution properties like mean and variance. It is particularly useful in large-scale applications because of its low computational overhead. However, it does not directly identify the cause of performance degradation or apply to categorical features.
- A. Correct.
Correct: Summary statistic monitoring involves comparing properties such as mean, standard deviation, or quantiles of numeric features between training and production datasets. This helps detect shifts in the data distribution.
- B. Correct.
Correct: Significant changes in summary statistics, like an increase in variance or a shift in mean, can signal potential feature drift that might impact model performance.
- C. Incorrect.
Incorrect: Summary statistic monitoring is not typically used for categorical features. Instead, frequency distributions or other specialized methods are more suitable for categorical data.
- D. Correct.
Correct: Monitoring summary statistics is computationally efficient as it involves simple calculations on numeric data, making it suitable for real-time or large-scale data pipelines.
- E. Incorrect.
Incorrect: While summary statistic monitoring can detect potential drift, it does not directly identify the root cause of performance issues. Further analysis is required to diagnose the specific reasons for degradation.