import numpy as np
from sklearn.datasets import make_low_rank_matrix

from decompose.sklearn import DECOMPOSE
from decompose.distributions.cenNormal import CenNormal


# create a numpy array containing a synthetic low rank dataset
X = make_low_rank_matrix(n_samples=1000, n_features=1000,
                         effective_rank=3, tail_strength=0.5)

# create an instance of a decompose model
model = DECOMPOSE(modelDirectory="/tmp/myNewModel",
                  priors=[CenNormal(), CenNormal()],
                  n_components=3)

# train the model and transform the training data
U0 = model.fit_transform(X)

# learned filter bank
U1 = model.components_

# variance ratio of the sources
varianceRatio = model.variance_ratio_

# reconstruction of the data
XHat = np.dot(U0.T, U1)