def plot_avg_psd_encoder_weights(agg, data_dir='/auto/tdrive/mschachter/data', decomp='full_psds'):
freqs,lags = get_freqs_and_lags()
bs_agg = AggregateBiosounds.load(os.path.join(data_dir, 'aggregate', 'biosound.h5'))
wdf,Wsq,Wsq_by_freq,Wsq_by_reg = get_encoder_weights_squared(agg, decomp)
print 'Wsq_by_freq.shape=',Wsq_by_freq.shape
Wsq_by_freq,aprop_order = reorder_by_row_sum(Wsq_by_freq)
figsize = (23, 13)
fig = plt.figure(figsize=figsize)
fig.subplots_adjust(top=0.95, bottom=0.02, right=0.99, left=0.05, hspace=0.25, wspace=0.25)
gs = plt.GridSpec(100, 100)
pwidth = 45
# plot the performance boxplots by frequency
freqs = list(sorted(wdf.freq.unique()))
ax = plt.subplot(gs[55:95, 62:98])
plot_lfp_and_spike_perf_boxplot(agg, ax, lfp_decomp=decomp)
"""
# plot the average encoder effect sizes
ax = plt.subplot(gs[45:100, :pwidth])
plt.imshow(Wsq_by_freq, origin='upper', interpolation='nearest', aspect='auto', vmin=0, cmap=magma)
plt.xticks(range(len(freqs)), ['%d' % int(f) for f in freqs])
aprop_lbls = [ACOUSTIC_PROP_NAMES[USED_ACOUSTIC_PROPS[k]] for k in aprop_order]
plt.yticks(range(len(aprop_lbls)), aprop_lbls)
plt.xlabel('LFP Frequency (Hz)')
plt.colorbar(label='Normalized Encoder Effect')
"""
fname = os.path.join(get_this_dir(), 'average_encoder_weights.svg')
plt.savefig(fname, facecolor='w', edgecolor='none')
plt.show()
def draw_joint_relationships(data_dir='/auto/tdrive/mschachter/data'):
big3_file = os.path.join(data_dir, 'aggregate', 'acoustic_big3.csv')
if not os.path.exists(big3_file):
bs_file = os.path.join(data_dir, 'aggregate', 'biosound.h5')
bs_agg = AggregateBiosounds.load(bs_file)
aprops = ['maxAmp', 'sal', 'meanspect']
Xz, good_indices = bs_agg.remove_duplicates(aprops=aprops, thresh=np.inf)
stim_types = bs_agg.df.stim_type[good_indices].values
pdata = {'maxAmp': Xz[:, 0], 'sal': Xz[:, 1], 'meanspect': Xz[:, 2], 'stim_type': stim_types}
df = pd.DataFrame(pdata)
df.to_csv(big3_file, header=True, index=False)
else:
df = pd.read_csv(big3_file)
print '# of syllables used: %d' % len(df)
plt.figure()
ax = plt.subplot(111, projection='3d')
maxAmp_rescaled = df.maxAmp
maxAmp_rescaled -= maxAmp_rescaled.min()
maxAmp_rescaled = np.log(maxAmp_rescaled + 1e-6)
maxAmp_rescaled -= maxAmp_rescaled.mean()
maxAmp_rescaled /= maxAmp_rescaled.std(ddof=1)
for ct in DECODER_CALL_TYPES:
i = df.stim_type == ct
if i.sum() == 0:
continue
x = maxAmp_rescaled[i]
y = df.sal[i].values
z = df.meanspect[i].values
c = CALL_TYPE_COLORS[ct]
ax.scatter(x, y, z, s=49, c=c)
ax.set_xlabel('Amplitude')
ax.set_ylabel('Saliency')
ax.set_zlabel('Mean Frequency')
leg = custom_legend([CALL_TYPE_COLORS[ct] for ct in DECODER_CALL_TYPES], DECODER_CALL_TYPES)
plt.legend(handles=leg)
plt.figure()
nrows = 1
ncols = 3
call_type_order = ['song', 'Te', 'DC', 'Be', 'LT', 'Ne', 'Ag', 'Di', 'Th']
# plot amplitude vs saliency
ax = plt.subplot(nrows, ncols, 1)
for ct in call_type_order:
i = df.stim_type == ct
if i.sum() == 0:
continue
x = maxAmp_rescaled[i]
y = df.sal[i].values
c = CALL_TYPE_COLORS[ct]
ax.scatter(x, y, s=49, c=c, alpha=0.7)
plt.xlabel('Maximum Amplitude')
plt.ylabel('Saliency')
plt.xlim(-3, 2)
# plot amplitude vs meanspect
ax = plt.subplot(nrows, ncols, 2)
for ct in call_type_order:
i = df.stim_type == ct
if i.sum() == 0:
continue
x = maxAmp_rescaled[i]
y = df.meanspect[i].values
c = CALL_TYPE_COLORS[ct]
ax.scatter(x, y, s=49, c=c, alpha=0.7)
plt.xlabel('Maximum Amplitude')
plt.ylabel('Mean Spectral Freq')
plt.xlim(-3, 2)
# plot saliency vs meanspect
ax = plt.subplot(nrows, ncols, 3)
for ct in call_type_order:
i = df.stim_type == ct
if i.sum() == 0:
continue
x = df.sal[i].values
y = df.meanspect[i].values
c = CALL_TYPE_COLORS[ct]
ax.scatter(x, y, s=49, c=c, alpha=0.7)
plt.xlabel('Saliency')
plt.ylabel('Mean Spectral Freq')
plt.show()
plot_spectrogram(sprops['spec_t'][si:ei], sprops['spec_freq'], sprops['spec'][:, si:ei], ax=ax,
colormap='SpectroColorMap', colorbar=False)
plt.title('%d_%d (%s), fund=%0.2f' % (stim_id, stim_order, stim_type, f))
fi = yes_fund_i[k]
f = funds[fi]
stim_id, stim_order, stim_type = stims[fi]
sprops = get_syllable_props(agg, stim_id, stim_order, data_dir)
ax = plt.subplot(gs[1, k])
si = sprops['spec_si']
ei = sprops['spec_ei']
plot_spectrogram(sprops['spec_t'][si:ei], sprops['spec_freq'], sprops['spec'][:, si:ei], ax=ax,
colormap='SpectroColorMap', colorbar=False)
plt.title('%d_%d (%s), fund=%0.2f' % (stim_id, stim_order, stim_type, f))
plt.show()
if __name__ == '__main__':
set_font()
agg_file = '/auto/tdrive/mschachter/data/aggregate/biosound.h5'
agg = AggregateBiosounds.load(agg_file)
# plot_syllable_comps(agg)
plot_acoustic_stats(agg)
# plot_nofund(agg)
def plot_avg_pairwise_encoder_weights(agg, data_dir='/auto/tdrive/mschachter/data'):
freqs,lags = get_freqs_and_lags()
bs_agg = AggregateBiosounds.load(os.path.join(data_dir, 'aggregate', 'biosound.h5'))
print agg.df.decomp.unique()
decomp = 'full_psds+full_cfs'
i = agg.df.decomp == decomp
assert i.sum() > 0
W = list()
wdata = {'lag':list(), 'xindex':list()}
nlags = len(lags)
g = agg.df[i].groupby(['bird', 'block', 'segment', 'hemi'])
for (bird,block,seg,hemi),gdf in g:
assert len(gdf) == 1
wkey = gdf['wkey'].values[0]
index2electrode = agg.index2electrode[wkey]
eperf = agg.encoder_perfs[wkey]
eweights = agg.encoder_weights[wkey]
# site = '%s_%s_%s_%s' % (bird, block, seg, hemi)
for k,e1 in enumerate(index2electrode):
for j in range(k):
for l,lag in enumerate(lags):
w = eweights[k, j, l, :]
wdata['lag'].append(int(lag))
wdata['xindex'].append(len(W))
W.append(w)
wdf = pd.DataFrame(wdata)
W = np.array(W)
W[np.isnan(W)] = 0.
W2 = W**2
print 'nlags=%d' % nlags
W2_by_lag = np.zeros([len(USED_ACOUSTIC_PROPS), nlags])
for l,lag in enumerate(lags):
i = wdf.lag == int(lag)
assert i.sum() > 0
ii = wdf.xindex[i].values
W2_by_lag[:, l] = W2[ii, :].mean(axis=0)
plot = True
if plot:
plt.figure()
# ax = plt.subplot(1, 3, 1)
absmax = W2_by_lag.max()
plt.imshow(W2_by_lag, interpolation='nearest', aspect='auto', cmap=magma, vmin=0, vmax=absmax)
plt.colorbar()
plt.title('Mean Weights')
plt.yticks(np.arange(len(USED_ACOUSTIC_PROPS)), USED_ACOUSTIC_PROPS)
plt.xticks(np.arange(nlags), ['%d' % x for x in lags])
plt.show()
return lags,W2_by_lag,None
