The tools that get data in and ready.
Everything an agent needs before it can train: bringing a file into TuiML, looking at it, reshaping it, or synthesising one when there is no real data to hand.
Tools
-
tuiml_upload_data: Register a dataset from a file path or from inline
text, giving it a name later tools can refer to.
-
tuiml_read_data: Show rows (head, tail, sample, or specific indices)
so the agent can see what it is working with.
-
tuiml_profile_data: Shape, dtypes, missing values, class balance and
summary statistics.
-
tuiml_generate_data: Synthesise a dataset from a built-in generator
(Blobs, Agrawal, Friedman, ...).
-
tuiml_preprocess: Apply preprocessing steps and save the result.
-
tuiml_select_features: Reduce to the most informative columns.
Notes
Anywhere these take a data argument, it accepts a built-in dataset name ("iris"), a path to a file, or the name of a previous upload.