#ax.set_yticks([])
fig.tight_layout()
#ax.set_frame_on(False)
# do left hand side table showing
J = 3
K = 4
D = 11
N = 25
X, true_theta = utils.generate_data(n_samples=N,
n_features=D,
n_components=K,
n_latent_factors=J,
omega_scale=0.1,
noise_scale=0.1,
random_seed=random_seed)
scores = true_theta["scores"]
tau = true_theta["R"]
def draw_matrix(ax,
x,
y,
w,
h,
horizontal_grid_spacing=1,
vertical_grid_spacing=1,
data_kwds = dict(n_features=n_features,
n_components=n_components,
n_latent_factors=n_latent_factors,
n_samples=n_samples,
omega_scale=omega_scale,
noise_scale=noise_scale,
random_seed=random_seed)
mcfa_kwds = dict(tol=1e-5,
max_iter=10000,
init_factors="random",
init_components="random",
random_seed=random_seed,
covariance_regularization=1e-6)
Y, truth = utils.generate_data(**data_kwds)
truth_packed = (truth["pi"], truth["A"], truth["xi"], truth["omega"],
truth["psi"])
for missing_data_fraction in (0, 0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5, 0.75,
0.9):
# Throw away some data.
missing_flat_indices = np.random.choice(Y.size,
int(missing_data_fraction *
Y.size),
replace=False)
is_missing = np.zeros(Y.size, dtype=bool)
is_missing[missing_flat_indices] = True
is_missing = is_missing.reshape(Y.shape)
