latents 0.0.5

A library for latent variable modeling and dimensionality reduction

Latents: A Python Library for Latent Variable Modeling

[!WARNING] Latents is in an early stage of development and is not yet stable. The API is subject to frequent change.

Latents is a Python library for latent variable modeling and dimensionality reduction, with an emphasis on linear, probabilistic methods. The following methods are currently supported:

• Group Factor Analysis (GFA)
• Delayed Latents Across Multiple Groups (mDLAG) — under development

Installation

Requires Python 3.10 or higher.

pip install latents

Quickstart

from latents.callbacks import ProgressCallback
from latents.gfa import GFAFitConfig, GFAModel

# Configure and fit to multi-group observation data Y
config = GFAFitConfig(x_dim_init=10)  # Start with more dims than needed
model = GFAModel(config=config)
model.fit(Y, callbacks=[ProgressCallback()])

# Inspect results
model.flags.display()

# Discovered sparsity pattern: which factors are significant in each group
_, sig_dims, _, _ = model.obs_posterior.compute_dimensionalities()
print(sig_dims.astype(int))

Documentation

See our documentation for the complete API reference, user guide, and tutorials.

Contributing

Interested in contributing? See the Contributing Guide.

Code of Conduct

Please consult the Code of Conduct.

License

MIT