Snowflake SPS-C01 real exam prep : Snowflake Certified SnowPro Specialty - Snowpark

You are developing a Snowpark Python application that reads data from a Snowflake table, performs several transformations including filtering, aggregation, and joining with another DataFrame, and then writes the results back to a new table. You want to optimize the execution plan to minimize data movement and processing time. Which of the following strategies would be MOST effective in leveraging Snowpark's lazy evaluation capabilities to achieve this optimization?

• A. Executing each transformation in separate Python processes using multiprocessing to parallelize the workload.
• B. Calling 'cache()' on the initial DataFrame read from the table to materialize it in memory before any transformations.
• C. Chaining all the transformations together using DataFrame methods (e.g., 'filter()' , 'groupBy()' , 'join()') and only calling or at the very end.
• D. Calling after each transformation to materialize intermediate results and then creating new DataFrames for subsequent operations.
• E. Defining all transformations in a single, complex SQL query string and using to execute it.

You have developed a Snowpark Python stored procedure that calculates the average sales per region from a large sales data table. The procedure is currently defined inline within your Snowflake notebook. You want to operationalize this by creating the stored procedure from a local Python file named The file contains the following code: "'python from snowflake.snowpark.session import Session def calculate_avg_sales(session: Session, sales_table_name: str, region_column: str, sales_column: str) -> float: sales df = session.table(sales table name) avg_sales df = sales_df.group_by(region_column).agg({sales_column: 'avg'}) avg_sales = avg_sales_df.collect() return avg_sales[0][1] Which of the following code snippets correctly creates the stored procedure 'AVG SALES PROC in Snowflake, referencing the Python file, and handles potential dependency issues? Assume you have already established a Snowpark session named 'session' and that the stage 'my_stage' already exists.

• A.
  
• B.
  
• C.
  
• D.
  
• E.
  

You need to perform a set difference operation between two DataFrames in Snowpark Python. 'dfl' contains customer IDs from a marketing campaign, and 'df2 contains customer IDs from a recent purchase event. You want to identify customers who were targeted in the campaign but did not make a recent purchase. Both DataFrames have a column named 'customer id'. Which of the following approaches provides the most efficient way to accomplish this task in Snowpark?

• A.
  
• B.
  
• C.
  
• D.
  
• E.
  

You have a Snowpark DataFrame named with columns 'category', , and You want to perform the following transformations using Snowpark:

• A.
  
• B.
  
• C.
  
• D.
  
• E.
  

Consider a DataFrame 'products df loaded from a SnoMlake table. It contains a 'features' column of type VARIANT, where each row contains a JSON object representing product features. Your task is to create a new DataFrame where each feature becomes a separate column. You need to dynamically extract these features without knowing the specific feature names in advance. Which of the following approaches could achieve this using Snowpark, and what considerations are important? Choose all that apply:

• A. Use a User-Defined Function (UDF) to parse the JSON and return a dictionary. Then, use a loop to iterate over the keys in the dictionary and create new columns based on these keys.
• B. It's not possible to dynamically extract feature names and create columns in Snowpark without knowing the schema in advance.
• C. The function can be used in conjunction with a Snowpark SQL query to dynamically extract the json into separate columns.
• D. Use the 'FLATTEN' function within a Snowpark DataFrame transformation. This allows you to transform the key-value pairs within the VARIANT column into separate rows, which can then be pivoted to create new columns.
• E. Use the function on the VARIANT column to get an array of feature names. Then, use a loop to iterate over this array and dynamically create new columns using bracket notation (e.g.,