Why is feature scaling (e.g., standardization) often applied before training a linear model?

To convert categorical variables into numeric ones

To ensure features contribute equally to the model and improve convergence

To handle missing values in the dataset

To reduce the number of features in the dataset

A data scientist is predicting whether a user will click an ad. The dataset has 'time_of_day' (hour 0-23), 'user_age', and 'ad_category'. The scientist suspects that the effect of 'time_of_day' on click probability depends on 'ad_category'. Which feature should be created to test this hypothesis?

Create an interaction feature 'time_of_day * ad_category_encoded'.

Create a feature 'is_weekend' from 'time_of_day'.

Bin 'time_of_day' into morning, afternoon, evening, and night.

Create a feature 'user_age / time_of_day'.

A data scientist is preparing a dataset with a high-cardinality categorical column 'postal_code' containing over 10,000 unique values for a gradient boosting model. The goal is to retain the predictive signal of the column while keeping the feature space manageable. Which encoding approach is most appropriate?

Drop the column entirely to avoid the curse of dimensionality.

Apply target encoding with cross-validation to avoid overfitting.

Apply one-hot encoding and rely on the model's feature importance to prune irrelevant columns.

Apply label encoding and treat the values as ordinal integers.

Why is feature scaling particularly important for linear models but less critical for tree-based models?

Tree models are always more accurate than linear models

Linear models use distance or gradient-based optimization, while trees split on thresholds and are scale-invariant

Tree models require all features to be normally distributed

Linear models cannot handle categorical variables

Which of the following operations creates an interaction feature between two numeric variables X and Y?

Standardizing X and Y separately

SnowflakeSnowPro Advanced — Data Scientist

Domain 3Objective 2

Engineer Features for Models SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 7)

Part of the Data Preparation and Feature Engineering domain, which makes up ~26% of our current practice bank.

45questions here

9free pages

15concepts

Questions 31–35

31foundation · easy
Why is feature scaling (e.g., standardization) often applied before training a linear model?
Select an answer first
32application · medium
A data scientist is predicting whether a user will click an ad. The dataset has 'time_of_day' (hour 0-23), 'user_age', and 'ad_category'. The scientist suspects that the effect of 'time_of_day' on click probability depends on 'ad_category'. Which feature should be created to test this hypothesis?
Select an answer first
33application · medium
A data scientist is preparing a dataset with a high-cardinality categorical column 'postal_code' containing over 10,000 unique values for a gradient boosting model. The goal is to retain the predictive signal of the column while keeping the feature space manageable. Which encoding approach is most appropriate?
Select an answer first
34foundation · easy
Why is feature scaling particularly important for linear models but less critical for tree-based models?
Select an answer first
35foundation · easy
Which of the following operations creates an interaction feature between two numeric variables X and Y?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to SNOWPRO-ADVANCED-DATA-SCIENTIST

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Snowflake. “SNOWPRO-ADVANCED-DATA-SCIENTIST” is a trademark of its owner, used for identification only.