GCP Professional Data Engineer Practice Question

Your team is building a BigQuery ML regression model with CREATE MODEL and a TRANSFORM clause. Column promotion_code is a STRING with roughly 40 unique values. Column order_total is a NUMERIC field that contains many extreme outliers. You must 1) convert promotion_code into individual binary indicator features and 2) rescale order_total so that the model is less sensitive to outliers without clipping zeros. Which combination of manual preprocessing functions satisfies both requirements?

Use ML.ONE_HOT_ENCODER on promotion_code and ML.MAX_ABS_SCALER on order_total
Use ML.BUCKETIZE on promotion_code and ML.ROBUST_SCALER on order_total
Use ML.FEATURE_CROSS on promotion_code and ML.NORMALIZER on order_total
Use ML.ONE_HOT_ENCODER on promotion_code and ML.ROBUST_SCALER on order_total

Report Issue

Answer Description

ML.ONE_HOT_ENCODER transforms each distinct string value in promotion_code into a separate binary feature, which is the standard way to represent low-cardinality categorical variables for linear models. ML.ROBUST_SCALER scales a numeric column by subtracting its median and dividing by the inter-quartile range, making the transformation resistant to extreme outliers while retaining the original zero point if the median is near zero. The other options either apply a scaler that is not robust to outliers, use bucketization or feature crossing instead of one-hot encoding, or normalize the numeric column in a way that remains sensitive to outliers.

Ask Bash

Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.