Snowflake Certified SnowPro Specialty - Snowpark Sample Questions:

1. You have a Snowpark DataFrame containing product information, and you want to persist it into a Snowflake table named PRODUCTS. You need to handle the following scenarios: 1. If the table 'PRODUCTS does not exist, create it. 2. If the table PRODUCTS' exists, append the data from 'df_products' to it. Which of the following methods can achieve this?

A)

B)

C)

D)

E)

2. You are working with a Snowpark DataFrame named 'sales df' containing sales data including columns 'product id', 'sale date', and 'sale_amount'. You want to create a new DataFrame 'filtered df that only includes rows where the 'product id' is present in a list of approved product IDs and the 'sale_amount' is greater than the average sale amount. You have already calculated the average sale amount and stored it in a variable named 'avg_sale amount'. Which of the following code snippets correctly achieves this?

A)

B)

C)

D)

E)

3. You are developing a Snowpark application that performs feature engineering on a dataset of customer transactions. This involves calculating several complex aggregate features such as rolling averages, medians, and custom ratios. You want to optimize the performance of this feature engineering process using a Snowpark-optimized warehouse. Which of the following strategies would be MOST effective in achieving optimal performance?

A) Implement all feature engineering calculations using Python User-Defined Functions (UDFs) and apply them to the Snowpark DataFrame.
B) Use stored procedures implemented in Java within Snowpark for feature calculations.
C) Leverage Snowpark's built-in functions and SQL expressions as much as possible for feature engineering, and rewrite performance-critical calculations as Java or Scala User-Defined Table Functions (UDTFs).
D) Materialize intermediate Snowpark DataFrames after each feature engineering step to avoid recomputation.
E) Use the 'GROUP BY clause in Snowpark SQL to compute aggregate features, leveraging window functions where appropriate for rolling calculations.

4. You are optimizing a Snowpark Python application that performs complex data transformations on a large dataset. You notice significant performance bottlenecks. Which of the following optimization techniques would be MOST effective in leveraging the Snowpark architecture to improve performance?

A) Exploiting lazy evaluation by chaining transformations together and avoiding unnecessary 'collect()' or 'toPandas()' calls.
B) Manually partitioning the DataFrame into smaller chunks before applying transformations.
C) Using 'session.sql()' whenever possible instead of Snowpark DataFrame operations.
D) Using User-Defined Functions (UDFs) written in Python for all transformations, ensuring they are vectorized where possible, instead of native Snowpark functions.
E) Converting all Snowpark DataFrames to Pandas DataFrames before performing any transformations.

5. You are developing a Snowpark application that involves creating a set of stored procedures and UDFs to process data'. To ensure proper version control and dependency management, you decide to package your Python code into a single Python Wheel file and deploy it to Snowflake. Which of the following methods are valid for deploying and utilizing this Python Wheel file within Snowflake, considering best practices for maintainability and security? (Select TWO)

A) Use the 'snowflake-cli' to push the Python Wheel file as a package, then add the package name to the list of packages when creating the stored procedure or UDF.
B) Use the Snowsight UI to upload the Python Wheel file as a dependency for the Snowflake environment, making it available for all stored procedures and UDFs within that environment.
C) Upload the Python Wheel file to an internal stage and directly reference it in the UDF or stored procedure definition using the 'USING' clause.
D) Create a Conda environment file (environment.yml) that specifies the Python Wheel file as a dependency and use this file to create a Snowflake environment. Then, associate the stored procedures and UDFs with that environment.
E) Upload the Python Wheel file to an external stage (e.g., AWS S3) and configure Snowflake to access the external stage, then reference the wheel file path in the USING' clause of the UDF or stored procedure.

Solutions: