def plot_preds(cellid, results, source, fname):
fig = plt.figure(figsize=(18, 9))
fig.set_facecolor('white')
plt.suptitle('Experiment: %s: %s Source' % (cellid, source))
cols = len(results)
legend = True
for i, [t, dat] in enumerate(results.iteritems()):
#dat = results[t]
pred = dat['pred']
mn = dat['mn']
std = dat['std']
crr_pred = dat['crr_pred']
crr_exp = dat['crr_exp']
plot_mean_std(None, mn, std, '%s - Corr Trials to Mean: %.2f' %
(t, crr_exp), [2, cols, 1 + i], legend=legend)
plot_prediction(pred, mn, 'Prediction Corr to Mean: %.2f' % (crr_pred),
[2, cols, cols + 1 + i], legend=legend)
legend = False
plt.subplots_adjust(left=0.05, bottom=0.06, right=0.97,
top=0.87, wspace=0.3, hspace=0.6)
#plt.show()
fig.savefig(fname + '.eps')
fig.savefig(fname + '.png')
plt.close(fig)
def plot_preds_fourier(cellid, results, fname):
fig = plt.figure(figsize=(18, 9))
fig.set_facecolor('white')
plt.suptitle('Experiment: %s Fourier Source' % (cellid))
cols = len(results)
legend = True
for i, [t, dat] in enumerate(results.iteritems()):
#dat = results[t]
pred = dat['pred']
mn = dat['mn']
std = dat['mn']
crr_pred = dat['crr_pred']
crr_exp = dat['crr_exp']
weights = dat['coefs'].mean(0)
plot_params = dat['plot_params']
plot_mean_std(None,
mn,
std,
'%s - Corr Trials to Mean: %.2f' % (t, crr_exp),
[3, cols, 1 + i],
legend=legend)
plot_prediction(pred,
mn,
'Prediction Corr to Mean: %.2f' % (crr_pred),
[3, cols, cols + 1 + i],
legend=legend)
legend = False
all_weights = None
inds = [0, 0]
for f in plot_params['idx_four']:
four_weights = weights[f].reshape(plot_params['four_shape'][0])
ff = reverse_fft(four_weights)
if all_weights is None:
nx = int(np.sqrt(len(plot_params['idx_four'])))
x = plot_params['four_shape'][0][0]
xx = x * nx
all_weights = np.zeros([xx, xx])
all_weights[inds[0]:inds[0] + x, inds[1]:inds[1] + x] = ff
inds[1] += x
if inds[1] == xx:
inds[0] += x
inds[1] = 0
plot_four(all_weights, [0, all_weights.max()],
[3, cols, 2 * cols + 1 + i],
interpolation='bilinear')
plt.subplots_adjust(left=0.05,
bottom=0.06,
right=0.97,
top=0.87,
wspace=0.3,
hspace=0.6)
#plt.show()
fig.savefig(fname + '.eps')
fig.savefig(fname + '.png')
plt.close(fig)
def plot_preds_fourier(cellid, results, fname):
fig = plt.figure(figsize=(18, 9))
fig.set_facecolor('white')
plt.suptitle('Experiment: %s Fourier Source' % (cellid))
cols = len(results)
legend = True
for i, [t, dat] in enumerate(results.iteritems()):
#dat = results[t]
pred = dat['pred']
mn = dat['mn']
std = dat['mn']
crr_pred = dat['crr_pred']
crr_exp = dat['crr_exp']
weights = dat['coefs'].mean(0)
plot_params = dat['plot_params']
plot_mean_std(None, mn, std, '%s - Corr Trials to Mean: %.2f' %
(t, crr_exp), [3, cols, 1 + i], legend=legend)
plot_prediction(pred, mn, 'Prediction Corr to Mean: %.2f' % (crr_pred),
[3, cols, cols + 1 + i], legend=legend)
legend = False
all_weights = None
inds = [0, 0]
for f in plot_params['idx_four']:
four_weights = weights[f].reshape(plot_params['four_shape'][0])
ff = reverse_fft(four_weights)
if all_weights is None:
nx = int(np.sqrt(len(plot_params['idx_four'])))
x = plot_params['four_shape'][0][0]
xx = x * nx
all_weights = np.zeros([xx, xx])
all_weights[inds[0]: inds[0] + x,
inds[1]: inds[1] + x] = ff
inds[1] += x
if inds[1] == xx:
inds[0] += x
inds[1] = 0
plot_four(all_weights, [0, all_weights.max()],
[3, cols, 2 * cols + 1 + i],
interpolation='bilinear')
plt.subplots_adjust(left=0.05, bottom=0.06, right=0.97,
top=0.87, wspace=0.3, hspace=0.6)
#plt.show()
fig.savefig(fname + '.eps')
fig.savefig(fname + '.png')
plt.close(fig)
def plot_preds(cellid, results, source, fname):
fig = plt.figure(figsize=(18, 9))
fig.set_facecolor('white')
plt.suptitle('Experiment: %s: %s Source' % (cellid, source))
cols = len(results)
legend = True
for i, [t, dat] in enumerate(results.iteritems()):
#dat = results[t]
pred = dat['pred']
mn = dat['mn']
std = dat['std']
crr_pred = dat['crr_pred']
crr_exp = dat['crr_exp']
plot_mean_std(None,
mn,
std,
'%s - Corr Trials to Mean: %.2f' % (t, crr_exp),
[2, cols, 1 + i],
legend=legend)
plot_prediction(pred,
mn,
'Prediction Corr to Mean: %.2f' % (crr_pred),
[2, cols, cols + 1 + i],
legend=legend)
legend = False
plt.subplots_adjust(left=0.05,
bottom=0.06,
right=0.97,
top=0.87,
wspace=0.3,
hspace=0.6)
#plt.show()
fig.savefig(fname + '.eps')
fig.savefig(fname + '.png')
plt.close(fig)
sigs.append(len(trl))
if len(trl) == 0:
trl = [0]
ys.append(np.mean(trl))
stds.append(np.std(trl))
xs = np.array(xs)
ys = np.array(ys)
stds = np.array(stds)
if (ys + stds).max() > mx:
mx = (ys + stds).max()
sigs = np.array(sigs)
plt.figure(fig1.number)
axs.append(
plot_mean_std(xs,
ys,
stds,
title, [a, b, cnt],
legend=False,
line='-o'))
plt.xlim(shifts.min() - 0.5, shifts.max() + 0.5)
plt.xticks(xaxis_time)
plt.figure(fig2.number)
plot_mean_std(xs,
sigs,
np.zeros_like(sigs),
title, [a, b, cnt],
legend=False,
line='-o')
plt.xlim(shifts.min() - 0.5, shifts.max() + 0.5)
plt.ylim(0, n)
plt.figure(fig1.number)
plt.subplots_adjust(left=0.03,
crr_sum[k][str(shift)] = [len(trl[trl != 0]),
trl.tolist()]
trl = trl[trl != 0]
sigs.append(len(trl))
if len(trl) == 0:
trl = [0]
ys.append(np.mean(trl))
stds.append(np.std(trl))
xs = np.array(xs)
ys = np.array(ys)
stds = np.array(stds)
if (ys + stds).max() > mx:
mx = (ys + stds).max()
sigs = np.array(sigs)
plt.figure(fig1.number)
axs.append(plot_mean_std(xs, ys, stds, title, [a, b, cnt],
legend=False, line='-o'))
plt.xlim(shifts.min() - 0.5, shifts.max() + 0.5)
plt.xticks(xaxis_time)
plt.figure(fig2.number)
plot_mean_std(xs, sigs, np.zeros_like(sigs), title, [a, b, cnt],
legend=False, line='-o')
plt.xlim(shifts.min() - 0.5, shifts.max() + 0.5)
plt.ylim(0, n)
plt.figure(fig1.number)
plt.subplots_adjust(left=0.03, bottom=0.05, right=0.97, top=0.95,
wspace=0.23, hspace=0.1)
stim_types = sorted(stim_types)
stim_types_short = []
for s in stim_types:
stim_types_short.append(s[:s.find('_')])
mx *= 1.1
