API Reference / cli /

generate.py

Generate Command - Create synthetic datasets for testing and demos.

Functions

Command

tuiml generate

Line 17
tuiml generate [OPTIONS]

Generate a synthetic dataset for testing and demos.

Writes a freshly generated dataset to a CSV file and reports its path, shape and a short preview, ready to feed straight into tuiml train. Classification generators (RandomRBF, Agrawal, LED, Hyperplane), regression generators (Friedman, MexicanHat, Sine) and clustering generators (Blobs, Moons, Circles, SwissRoll) are all supported. Pass --random-seed to make the output reproducible, and --generator-params for anything the dedicated options do not cover.

Generate a clustering dataset with three clusters:

bash 
$ tuiml generate --generator Blobs --n-samples 500 --n-clusters 3

Generate a reproducible classification dataset:

bash 
$ tuiml generate --generator RandomRBF --n-classes 4 --random-seed 42

Generate a noisy regression dataset:

bash 
$ tuiml generate --generator Friedman --n-samples 200 --noise 0.1

Options

--generator
str required
Generator class name (e.g., RandomRBF, Friedman, Blobs)
--n-samples
int
Number of samples to generate
--n-features
int
Number of features (not all generators support this)
--n-classes
int
Number of classes (classification generators only)
--n-clusters
int
Number of clusters (clustering generators only)
--noise
float
Noise level (regression generators only)
--random-seed
int
Random seed for reproducibility
--generator-params
str
Additional generator-specific parameters as a JSON object, e.g. '{"n_centroids": 20}'
--json-output
flag
Output raw JSON