def _event_explained(self, ev, padding=15):
"""check event for explanation by the filter bank"""
# early exit if no discriminants are present
if not self._disc.size:
return False
# cut relevant piece of the discriminants
data_ep = ev - self._learn_templates,\
ev - self._learn_templates + self.tf
disc_ep = data_ep[0] + self._tf / 2,\
data_ep[1] + self._tf / 2
if self.verbose.has_plot:
try:
from spikeplot import mcdata
ep = data_ep[0] - padding, disc_ep[1] + padding
mcdata(
data=self._chunk[ep[0]:ep[1]],
#other=self._disc[at[0]:at[1]], events=evts,
other=self._disc[ep[0]:ep[1]],
x_offset=ep[0],
events={0: [ev], 1: [data_ep[0] + self._tf]},
epochs={0: [data_ep], 1: [disc_ep]},
title='det@%s(%s) disc@%s' % (
ev, self._learn_templates, ev + self._tf),
show=True)
except ImportError:
pass
return self._disc[disc_ep[0] - padding:disc_ep[1] + padding,
:].max() >= 0.0
def plot(self, show=False):
"""plot detection in mcdata plot"""
try:
from spikeplot import plt, mcdata, COLOURS
except ImportError:
return None
fig = mcdata(self.data, other=self.energy, events={0: self.events},
show=False)
for i, th in enumerate(self.threshold):
fig.axes[-1].axhline(th, c=COLOURS[i % len(COLOURS)])
self._plot_additional(fig)
if show is True:
plt.show()
return fig
def plot_sorting(self, ph=None, show=False):
"""plot the sorting of the last data chunk
:type ph: plot handle
:param ph: plot handle top use for the plot
:type show: bool
:param show: if True, call plt.show()
"""
# get plotting tools
try:
from spikeplot import COLOURS, mcdata
except ImportError:
return None
# check
if self._data is None or self.rval is None or len(
self._idx_active_set) == 0:
warnings.warn('not initialised properly to plot a sorting!')
return None
# create events
ev = {}
if self.rval is not None:
temps = self.template_set
for i in self._idx_active_set:
if i in self.rval:
if self.rval[i].any():
ev[i] = (self.bank[i].xi, self.rval[i])
# create colours
cols = COLOURS[:self.nf]
# calc discriminants for single units
other = None
if self.nf > 0:
self.reset_history()
other = super(FilterBankSortingNode, self)._execute(self._data)
other += getattr(self, '_lpr_s', sp.log(1.e-6))
other -= [.5 * self.get_xcorrs_at(i)
for i in xrange(self.nf)]
# plot mcdata
return mcdata(self._data, other=other, events=ev,
plot_handle=ph, colours=cols, show=show)
try:
import unittest2 as ut
except ImportError:
import unittest as ut
from botmpy.common import mad_scaling
from botmpy.common.datafile import XpdFile
import scipy as sp
try:
from spikeplot import mcdata, plt
WITH_PLOT = True
except ImportError:
WITH_PLOT = False
if __name__ == '__main__':
xpd = XpdFile('/home/pmeier/Data/Munk/Louis/L014/L0140001.xpd')
data = xpd.get_data(item=1)
if WITH_PLOT:
mcdata(data, show=False)
data_scaled, scale = mad_scaling(data)
if WITH_PLOT:
mcdata(data_scaled, show=False)
print 'magic eq', sp.stats.norm.ppf(0.75)
print 'STD:', sp.std(data, 0), sp.std(data_scaled, 0)
if WITH_PLOT:
plt.show()
import scipy as sp
from spikeplot import mcdata
ns, nc = 300 + 1000, 4
tf = 65
cut = (-int(sp.floor(tf / 2.0)), int(sp.ceil(tf / 2.0)))
mydata = sp.randn(ns, nc)
wf = sp.array([sp.sin(sp.linspace(0, 2 * sp.pi, tf)) * (i + 1)
for i in xrange(nc)]).T
unit_ev = sp.array([50, 300, 750])
for i in unit_ev:
mydata[i + cut[0]:i + cut[1], :] += wf
myother = sp.randn(ns, 4)
ep = sp.array([[150, 200], [800, 950]])
ev = {0:(wf, unit_ev), 1:unit_ev}
mcdata(
mydata,
other=myother,
colours=None,
epochs=ep,
events=ev,
x_offset=-100)
