#plt.imshow(coeffs[stimclass,:,:])
#coeff_mag[stimclass] = np.linalg.norm(coeffs[stimclass,:,:],axis=1)
#Here, we'll CHOOSE a coefficient
coeff_mag[stimclass] = coeffs[stimclass,:,choose_band]
#Here, we'll do PCA to find the components across all bands
pca = PCA()
pca.fit(coeffs[stimclass,:,:])
rot_coeff = pca.fit_transform(coeffs[stimclass,:,:])
# plt.figure()
# for bb in range(5):
# kde = KernelDensity(kernel='gaussian', bandwidth=0.2).fit(rot_coeff[:,bb].reshape(-1,1))
# plt.plot(kde.score_samples(rot_coeff[:,bb].reshape(-1,1)))
# #plt.hist(rot_coeff[:,:])
#THIS ASSUMES COEFFICIENTS ARE ALL SIMILARLY DISTRIBUTED for each band
coeff_max = np.linalg.norm(np.abs(rot_coeff),axis=1)
coeff_max = stats.zscore(coeff_max)
alpha_coeffs = stats.zscore(np.abs(rot_coeff[:,2]))
mainfig = plt.figure()
EEG_Viz.plot_3d_scalp(alpha_coeffs>2,mainfig,clims=(0,3),animate=False,label=label_map[stimclass],unwrap=True)
plt.title(label_map[stimclass])
#%%
# need to do rPCA here for coefficients
_, _, SG = sig.spectrogram(sel_sig,
nperseg=512,
noverlap=500,
window=sig.get_window(
'blackmanharris', 512),
fs=fs / ds_fact)
ch_blip.append(np.max(SG[F_idxs[0], :]) - np.min(SG[F_idxs[0], :]))
ch_blip = np.array(ch_blip)
#%%
thresh = 0
ch_blip_z = stats.zscore(ch_blip)
plt.hist(ch_blip_z, bins=50, range=(-1, 1))
#%%
EEG_Viz.maya_band_display(ch_blip_z > thresh)
print(np.where(ch_blip_z > thresh))
#%%
#take out the alpha band in each channel
alpha = np.zeros((257))
theta = np.zeros((257))
for ch in range(257):
Fidx = np.where(np.logical_and(F > 8, F <= 14))
alpha[ch] = np.sum(diff_PSD[ch, Fidx], axis=1)
Fidx = np.where(np.logical_and(F > 4, F <= 8))
theta[ch] = np.sum(diff_PSD[ch, Fidx], axis=1)
#%%
#Topoplot it
#egipos = mne.channels.read_montage('/tmp/GSN257.sfp')
blintv = (0, 9)
elif pt == '907':
blintv = (0, 20)
sEEG.calc_baseline(intv=blintv)
#%%
sEEG.label_segments()
sEEG.plot_segment_labels()
#%%
# This is for 906
# Intervals are in SECONDS
if pt == '906':
intv_list = {'L': (11, 31), 'R': (51, 71), 'BL': (91, 111)}
elif pt == '907':
intv_list = {'L': (11, 31), 'R': (51, 71), 'BL': (91, 111)}
intv_order = ['L', 'R', 'BL']
median_response = []
for voltage in intv_order:
median_response.append(
sEEG.median_response(intv=intv_list[voltage], do_plot=True))
plt.suptitle(voltage)
#%%
median_response = np.array(median_response)
EEG_Viz.plot_3d_scalp(np.mean(median_response, axis=0)[:, 2], unwrap=True)