Databricks Machine Learning Associate Question 443
Select 3You are tasked with scoring a machine learning model using features stored in a Databricks Feature Store table. The model was trained using features from the same feature store table. Which of the following steps are necessary to correctly score the model using the feature store?
- A
Load the feature store table using the Feature Store client.
- B
Join the raw input data with the feature store table to retrieve features used for scoring.
- C
Manually preprocess the raw data to match the feature store schema.
- D
Use the Feature Store API to generate a feature lookup to retrieve the required features.
- E
Directly pass the raw input data to the model without involving the Feature Store.
Show answer and explanation
Correct answers: A, B, D
Explanation
Scoring a model with features from a feature store involves retrieving the required features using the Feature Store API, joining them with raw input data, and ensuring the features are aligned with the model's expectations. This process avoids manual preprocessing and ensures consistency with the features used during training.
- A. Correct.
Correct. Loading the feature store table using the Feature Store client is required to retrieve features for scoring.
- B. Correct.
Correct. Joining the raw input data with the feature store table ensures the required features are aligned with the input data for scoring.
- C. Incorrect.
Incorrect. Manual preprocessing is not necessary because the Feature Store handles feature retrieval and alignment.
- D. Correct.
Correct. The Feature Store API provides a feature lookup mechanism to fetch the specific features required for scoring.
- E. Incorrect.
Incorrect. Raw input data cannot be directly passed to the model because the model expects specific features from the feature store.