def test_tga_inverse():
# Test that the projection of data can be inverted
rng = np.random.RandomState(0)
n, p = 50, 3
X = rng.randn(n, p) # spherical data
X[:, 1] *= .00001 # make middle component relatively small
X += [5, 4, 3] # make a large mean
# same check that we can find the original data from the transformed
# signal (since the data is almost of rank n_components)
tga = TGA(n_components=2).fit(X)
Y = tga.transform(X)
Y_inverse = tga.inverse_transform(Y)
assert_almost_equal(X, Y_inverse, decimal=3)
def test_tga_median():
# TGA on dense arrays
tga = TGA(n_components=2, random_state=1, centering='median')
X = iris.data
X_r = tga.fit(X).transform(X)
assert_equal(X_r.shape[1], 2)
tga = TGA(n_components=2, random_state=1, centering='median')
X_r2 = tga.fit_transform(X)
assert_array_almost_equal(X_r, X_r2)
tga = TGA(random_state=1, centering='median')
tga.fit(X)
X_r = tga.transform(X)
tga = TGA(random_state=1, centering='median')
X_r2 = tga.fit_transform(X)
assert_array_almost_equal(X_r, X_r2)
if __name__ == "__main__":
import sys
import glob
import matplotlib.pyplot as pl
use_data = str(sys.argv[1])
M, shape = bitmap_to_mat(glob.glob(use_data + "/*.bmp")[:2000:2])
print(M.shape)
tga = TGA(n_components=5, random_state=1)
start_time = time.time()
tga.fit(M)
print("fitted, time taken {0}s".format(time.time() - start_time))
start_time = time.time()
transformed = tga.transform(M)
L = tga.inverse_transform(transformed)
print('calculated L, time taken {0}s'.format(time.time() - start_time))
S = M - L
if not os.path.exists('results_tga'):
os.makedirs('results_tga')
directory = "results_tga/" + use_data
if not os.path.exists(directory):
os.makedirs(directory)
fig, axes = pl.subplots(1, 3, figsize=(10, 4))
fig.subplots_adjust(left=0, right=1, hspace=0, wspace=0.01)
for i in range(min(len(M), 500)):
do_plot(axes[0], M[i], shape)