Kernel Computation kernel_computation

Kernel computation is a feature that pushes all data queries to the kernel side instead of the client side (in browser). This feature is useful when you have a large dataset locally and want high performance.

kernel_computation is a boolean parameter that can be set to True or False. The default value is None which means it will automatically choose the best value based on the dataset size.

For now, kernel_computation is using DuckDB as the backend for the computation. DuckDB is a high-performance in-memory database that is designed for analytical queries. If you want to use pygwalker with even larger datasets that cannot be handled locally, you can use a remote databse as dataset parameter to connect to a remote computation service, like snowflake or clickhouse.

import pygwalker as pyg
 
pyg.walk(df, kernel_computation=True)