class ICAMapperTests(unittest.TestCase):
def setUp(self):
# data: 40 sample feature line in 2d space (40x2; samples x features)
samples = N.vstack([N.arange(40.) for i in range(2)]).T
samples -= samples.mean()
samples += N.random.normal(size=samples.shape, scale=0.1)
self.ndlin = Dataset(samples=samples, labels=1, chunks=1)
# data: 40 sample feature line in 50d space (40x50; samples x features)
samples = N.vstack([N.arange(40.) for i in range(50)]).T
samples -= samples.mean()
samples += N.random.normal(size=samples.shape, scale=0.1)
self.largefeat = Dataset(samples=samples, labels=1, chunks=1)
self.pm = ICAMapper()
def testSimpleICA(self):
# train
self.pm.train(self.ndlin)
self.failUnlessEqual(self.pm.proj.shape, (2, 2))
# now project data into ICA space
p = self.pm.forward(self.ndlin.samples)
self.failUnlessEqual(p.shape, (40, 2))
# check that the mapped data can be fully recovered by 'reverse()'
self.failUnless(N.abs(self.pm.reverse(p) - self.ndlin.samples).mean() \
< 0.0001)
def setUp(self):
# data: 40 sample feature line in 2d space (40x2; samples x features)
samples = N.vstack([N.arange(40.) for i in range(2)]).T
samples -= samples.mean()
samples += N.random.normal(size=samples.shape, scale=0.1)
self.ndlin = Dataset(samples=samples, labels=1, chunks=1)
# data: 40 sample feature line in 50d space (40x50; samples x features)
samples = N.vstack([N.arange(40.) for i in range(50)]).T
samples -= samples.mean()
samples += N.random.normal(size=samples.shape, scale=0.1)
self.largefeat = Dataset(samples=samples, labels=1, chunks=1)
self.pm = ICAMapper()
