def test_partitionmapper():
ds = give_data()
oep = OddEvenPartitioner()
parts = list(oep.generate(ds))
assert_equal(len(parts), 2)
for i, p in enumerate(parts):
assert_array_equal(p.sa['partitions'].unique, [1, 2])
assert_equal(p.a.partitions_set, i)
assert_equal(len(p), len(ds))
def generate_testing_datasets(specs):
# Lets permute upon each invocation of test, so we could possibly
# trigger some funny cases
nonbogus_pool = np.random.permutation([0, 1, 3, 5])
datasets = {}
# use a partitioner to flag odd/even samples as training and test
ttp = OddEvenPartitioner(space='train', count=1)
for kind, spec in specs.iteritems():
# set of univariate datasets
for nlabels in [ 2, 3, 4 ]:
basename = 'uni%d%s' % (nlabels, kind)
nonbogus_features = nonbogus_pool[:nlabels]
dataset = normal_feature_dataset(
nlabels=nlabels,
nonbogus_features=nonbogus_features,
**spec)
# full dataset
datasets[basename] = list(ttp.generate(dataset))[0]
# sample 3D
total = 2*spec['perlabel']
nchunks = spec['nchunks']
data = np.random.standard_normal(( total, 3, 6, 6 ))
labels = np.concatenate( ( np.repeat( 0, spec['perlabel'] ),
np.repeat( 1, spec['perlabel'] ) ) )
data[:, 1, 0, 0] += 2*labels # add some signal
chunks = np.asarray(range(nchunks)*(total/nchunks))
mask = np.ones((3, 6, 6), dtype='bool')
mask[0, 0, 0] = 0
mask[1, 3, 2] = 0
ds = Dataset.from_wizard(samples=data, targets=labels, chunks=chunks,
mask=mask, space='myspace')
# and to stress tests on manipulating sa/fa possibly containing
# attributes of dtype object
ds.sa['test_object'] = [['a'], [1, 2]] * (ds.nsamples/2)
datasets['3d%s' % kind] = ds
# some additional datasets
datasets['dumb2'] = dumb_feature_binary_dataset()
datasets['dumb'] = dumb_feature_dataset()
# dataset with few invariant features
_dsinv = dumb_feature_dataset()
_dsinv.samples = np.hstack((_dsinv.samples,
np.zeros((_dsinv.nsamples, 1)),
np.ones((_dsinv.nsamples, 1))))
datasets['dumbinv'] = _dsinv
# Datasets for regressions testing
datasets['sin_modulated'] = list(ttp.generate(multiple_chunks(sin_modulated, 4, 30, 1)))[0]
# use the same full for training
datasets['sin_modulated_train'] = datasets['sin_modulated']
datasets['sin_modulated_test'] = sin_modulated(30, 1, flat=True)
# simple signal for linear regressors
datasets['chirp_linear'] = multiple_chunks(chirp_linear, 6, 50, 10, 2, 0.3, 0.1)
datasets['chirp_linear_test'] = chirp_linear(20, 5, 2, 0.4, 0.1)
datasets['wr1996'] = multiple_chunks(wr1996, 4, 50)
datasets['wr1996_test'] = wr1996(50)
datasets['hollow'] = Dataset(HollowSamples((40,20)),
sa={'targets': np.tile(['one', 'two'], 20)})
return datasets
from mvpa.clfs.svm import LinearCSVMC
from mvpa.measures.base import CrossValidation
from mvpa.measures.searchlight import sphere_searchlight
from mvpa.testing.datasets import datasets
from mvpa.mappers.fx import mean_sample
"""For the sake of simplicity, let's use a small artificial dataset."""
# Lets just use our tiny 4D dataset from testing battery
dataset = datasets['3dlarge']
"""Now it only takes three lines for a searchlight analysis."""
# setup measure to be computed in each sphere (cross-validated
# generalization error on odd/even splits)
cv = CrossValidation(LinearCSVMC(), OddEvenPartitioner())
# setup searchlight with 2 voxels radius and measure configured above
sl = sphere_searchlight(cv, radius=2, space='myspace',
postproc=mean_sample())
# run searchlight on dataset
sl_map = sl(dataset)
print 'Best performing sphere error:', np.min(sl_map.samples)
"""
If this analysis is done on a fMRI dataset using `NiftiDataset` the resulting
searchlight map (`sl_map`) can be mapped back into the original dataspace
and viewed as a brain overlay. :ref:`Another example <example_searchlight>`
shows a typical application of this algorithm.
specs = {'large' : { 'perlabel': 99, 'nchunks': 11,
'nfeatures': 20, 'snr': 8 * snr_scale},
'medium' :{ 'perlabel': 24, 'nchunks': 6,
'nfeatures': 14, 'snr': 8 * snr_scale},
'small' : { 'perlabel': 12, 'nchunks': 4,
'nfeatures': 6, 'snr' : 14 * snr_scale} }
# Lets permute upon each invocation of test, so we could possibly
# trigger some funny cases
nonbogus_pool = np.random.permutation([0, 1, 3, 5])
datasets = {}
# use a partitioner to flag odd/even samples as training and test
ttp = OddEvenPartitioner(space='train', count=1)
for kind, spec in specs.iteritems():
# set of univariate datasets
for nlabels in [ 2, 3, 4 ]:
basename = 'uni%d%s' % (nlabels, kind)
nonbogus_features = nonbogus_pool[:nlabels]
dataset = normal_feature_dataset(
nlabels=nlabels,
nonbogus_features=nonbogus_features,
**spec)
# full dataset
datasets[basename] = list(ttp.generate(dataset))[0]
