Data Transformation with Snowpark Proficiency (Advanced Level)

Topics covered

• connect to Snowflake from Jupyter Notebook and create and query tables using Snowpark APIs
• construct a UDTF that normalizes denormalized JSON data
• convert Snowpark pandas and Snowpark DataFrame objects and contrast their behaviors
• create and query Snowflake tables using Snowpark
• create a Snowflake Notebook and use Snowpark pandas via the Modin plugin
• create a table with semi-structured JSON data and query it using Snowpark
• identify the uses of stored procedures and contrast them with UDFs, UDTFs, and UDAFs in Snowflake
• implement the end_partition and init functions in a UDTF to achieve stateful processing
• implement the Snowpark equivalents of SELECT, FROM, WHERE, GROUP BY, and ORDER BY on DataFrame objects
• implement UDAFs to use Python objects and objects of user-defined classes
• implement union, intersect, and difference operations and joins on Snowpark DataFrames
• make standard Anaconda libraries and custom Python code available within a Snowpark handler
• outline UDAFs and methods for a UDAF handler class
• outline UDFs, UDTFs, UDAFs, and stored procedures and compare and contrast them
• outline UDTFs and partitioning in Snowflake and methods for UDTF handler class implementation
• partition rows and sort within a partition using UDTFs
• perform data transformations equivalent to SQL queries with group_by and order_by clauses in Snowpark
• register and invoke stored procedures
• register and invoke UDAFs to perform aggregation operations
• register and invoke UDTFs
• register anonymous UDFs using the udf function and session.udf.register, and invoke them using call_udf
• register permanent UDFs using the @udf decorator and then invoke them from different sessions
• register UDFs from SQL and Python files
• use the Snowpark APIs to create views from any DataFrame object
• write a Python function using the Snowpark APIs and then directly deploy it to a stored procedure