Missing value imputation with probabilistic models in Jax. Supported models are PCA, probabilistic PCA, and factor model.
$ git clone git@github.com:kstoneriv3/imputax.git
$ cd imputax
$ pip install -e .>>> import jax.numpy as jnp
>>> from imputax import impute_by_ppca
>>> X = jnp.array([[1, 2, 3], [2, None, 4], [None, 5, 6], [None, 6, None]], dtype=float)
>>> impute_by_ppca(X)
>>> impute_by_ppca(X, n_components=1)
Array([[1. , 2. , 3. ],
[2. , 2.4348595, 4. ],
[5.086713 , 5. , 6. ],
[5.0619698, 6. , 5.031307 ]], dtype=float32)
>>> Other available functions are impute_by_pca(...) and impute_by_factor_model(...).