See here for the example code that generated this page

Dataset Selection with Parquet Format

This example demonstrates the dataset selection feature combined with Parquet data format.

Key Features Demonstrated:

• Struct with DataFrames: A struct ExampleFrame with fields aa::DataFrame and bb::Union{Missing,DataFrame}
• Automatic extraction: DataFrames are extracted with dot-prefixed names (Symbol("struct_data.aa"), Symbol("struct_data.bb"))
• Subfolder storage: Struct DataFrames are stored in subfolders (e.g., data/struct_data/aa.parquet)
• Dataset switching: Use the dropdown to switch between datasets - data is loaded on-demand

How Parquet Storage Works:

When using dataformat = :parquet, DataFrames from structs are saved in subfolders:

• data/struct_data/aa.parquet - from example_struct.aa
• data/struct_data/bb.parquet - from example_struct.bb
• data/standalone_data.parquet - standalone DataFrame (no subfolder)

This is efficient for large datasets as data is loaded only when selected.