About the Exam

SnowPro Advanced: Data Scientist validates advanced knowledge and skills used to apply comprehensive data science principles, tools, and methodologies using Snowflake. It is aimed at data scientists with 2 or more years of hands-on experience using Snowflake in a production environment, and successful candidates may also have experience with Python, R, SQL, or PySpark. Passing demonstrates the ability to outline data science concepts, implement Snowflake best practices, prepare data and use feature engineering, train machine learning models, and use GenAI and LLM capabilities in Snowflake.

Exam Topics

Data science concepts25%
Implement Snowflake data science best practices20%
Prepare data and use feature engineering in Snowflake25%
Train and use machine learning models15%
Use GenAI and LLM capabilities in Snowflake15%

How to Use This Practice Exam

Browse — Read each question, select your answer, and reveal the explanation.
Exam Mode — Simulate real exam conditions with a timed session and score report.
Learn Mode — Spaced repetition schedules questions you struggle with for long-term retention.

Download the Full Exam PDF

Get every question and answer in a clean, printable PDF built for offline study. Purchase once, keep permanent access, and re-download the latest version anytime.

Get PDF

Learn More

Last updated May 14, 2026 at 2:12 AM

Topic filter

Retired questions

Question sort

Questions per page

QuestionQ1

Train and use machine learning models

A Data Scientist needs to obtain the coefficient of determination for two columns in a table.

Which function can be used to do this?

Explanation

REGR_R2(y, x) returns the coefficient of determination (R-squared) for non-null pairs in a group.

Learn more

REGR_R2 | Snowflake Documentation

Community Discussion

No comments yet. Be the first to start the discussion!

QuestionQ2

Use GenAI and LLM capabilities in Snowflake

A Data Scientist observed high usage of Snowflake Cortex LLM functions.

Which command restricts user access?

Explanation

SNOWFLAKE.CORTEX_USER is a Snowflake database role that is granted to the PUBLIC role by default, allowing users to access Cortex AI functions. Revoking that database role from PUBLIC removes the default broad access; the appropriate syntax is REVOKE DATABASE ROLE ... FROM ROLE PUBLIC.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!

QuestionQ3

Use GenAI and LLM capabilities in Snowflake

A Data Scientist needs to obtain a summary for every customer review. The data is held in a Snowflake table in a column named reviews.

How should this be accomplished with the LEAST operational overhead?

Explanation

SNOWFLAKE.CORTEX.SUMMARIZE is Snowflake’s managed SQL AI function for producing a summary of supplied English-language text, so it can be applied directly to each value in the reviews column. Embeddings are vector representations rather than summaries, while invoking a general completion model or deploying an LLM requires additional prompting or model-serving operations.

Learn more

SUMMARIZE (SNOWFLAKE.CORTEX)

Community Discussion

No comments yet. Be the first to start the discussion!

QuestionQ4

Train and use machine learning models

A Data Scientist trained a neural network for 50 epochs. After the network was trained, the results showed high accuracy on the training data but fairly low accuracy on the test data (for example, an overfitting problem).

Which techniques would help resolve the overfitting issue?

Choose two

Explanation

Regularization constrains model complexity, reducing the tendency to fit noise in the training set. Early stopping limits training once validation performance no longer improves, helping the model generalize better to unseen test data.

Community Discussion

No comments yet. Be the first to start the discussion!

QuestionQ5

Data science concepts

A Data Scientist observes that a model is generating more accurate predictions.

Which metric should be used to assess why the model’s predictions have improved?

Explanation

SHAP explains model predictions by assigning feature-attribution values, enabling analysis of which input features and feature effects contributed to the improvement in prediction accuracy.

Community Discussion

No comments yet. Be the first to start the discussion!

QuestionQ6

Train and use machine learning models

A Data Scientist wants to identify customer profiles that are likely to churn for a company. The training dataset contains 2% churn and 98% no churn, with churn encoded as 1 and no churn encoded as 0. The Data Scientist then ran the model in Snowpark.

After obtaining the confusion matrix, which value could give a misleading indication of model performance because the dataset is imbalanced?

Explanation

Accuracy can appear very high in a dataset with only 2% churn cases even when the model predicts every customer as no churn and identifies no actual churners. It is therefore not a reliable standalone performance measure for this imbalanced classification task.