def sensor_net(fiff, picks=None, name='fiff-sensors'):
"""
returns a sensor_net object based on the info in a fiff file.
"""
info = fiff.info
if picks is None:
chs = info['chs']
else:
chs = [info['chs'][i] for i in picks]
ch_locs = []
ch_names = []
for ch in chs:
x, y, z = ch['loc'][:3]
ch_name = ch['ch_name']
ch_locs.append((x, y, z))
ch_names.append(ch_name)
return _sensors.sensor_net(ch_locs, ch_names, name=name)
"""
Create simulated data with shape
(2 conditions * 15 subjects, 5 sensors, len(T) time points)
"""
# create the time dimension
T = var(np.arange(-.2, .8, .01), name='time')
# random data
x = np.random.normal(0,1,(30,5,len(T)))
# add an effect to the random data
x[15:,:3,20:40] += np.hanning(20) * 2
# create the sensor dimension from 5 sensor locations in 3d space
sensor = sensor_net([[0,0,0],[1,0,0],[0,-1,0],[-1,0,0],[0,1,0]],
name='testnet', transform_2d=None)
# combine all these into the ndvar
Y = ndvar(x, dims=('case', sensor, T), name='Y',
properties={'samplingrate':100, 'colorspace':cs.get_MEG(2)})
# To describe the cases ('case' dimension), create a condition and a subject factor
A = factor(['a0', 'a1'], rep=15, name='A')
subject = factor(xrange(15), tile=2, random=True, name='subject')
if 01: # plot topographically an uncorrected t-test
plot.topo.array(testnd.ttest(Y, A, match=subject))