Status: To Do
Fix Version/s: None
Component/s: Science Platform
This is a reminder-to-self ticket. The task is, at some point, to do some brainstorming about whether we can provide general means for performing reproducible sub-selections of tabular data results in the context of the Science Platform.
This is useful for purposes like the division of data into training, validation, and testing samples for machine learning, and the decimation of large results for "quick look" display.
It is often possible to implement such selections based on a hash. I would like to think about what it would take to do this in a way that was generically applicable and that could be connected to a simple UI.
E.g., I'd like to have an API, and a UI above it, that would enable actions like "divide the result into thirds" - reproducibly, so that the same division would be available in the Portal and programmatically in the Notebook or remotely.
(This is not something I want to spend any time on right now, however.)