for pos in range(length):
pvar[pos] = np.std(d[pos * step:(pos * step) + rslt, s])
#print p
lsignalsvar.append((pvar, datainfo.sensors[s]))
plt.subplots(figsize=(20, 10))
plt.axis([0, length, 0, 0.2])
plot(range(length), pvar)
#plt.show()
plt.title(datainfo.datafiles[dfile] + '-' +
datainfo.expnames[dfile] + '-' + datainfo.sensors[s],
fontsize=48)
plt.savefig(
datainfo.dpath + '/' + datainfo.name + '/Results/' +
'variance-' + datainfo.datafiles[dfile] + '-' +
datainfo.expnames[dfile] + '-' + datainfo.sensors[s] +
'.pdf',
orientation='landscape',
format='pdf')
plt.close()
plotSignals(
lsignalsvar,
6,
2,
0.2,
0,
'variance-' + datainfo.datafiles[dfile] + '-' +
datainfo.expnames[dfile] + '-' + str(rslt) + '-' + str(step),
datainfo.datafiles[dfile] + '-' + datainfo.expnames[dfile],
datainfo.dpath + '/' + datainfo.name + '/Results/',
orientation='portrait',
cstd=[0.05] * 12)
rects = ax.bar(ind+(i*width), h, width, color=colors[i])
fig.suptitle(datainfo.name + '-' + s + ext, fontsize=48)
minaxis = np.min(km.cluster_centers_)
maxaxis = np.max(km.cluster_centers_)
for nc in range(nclusters):
ax2 = fig.add_subplot(2, nclusters, nc+nclusters+1)
signal = km.cluster_centers_[lmax[nc][0]]
plt.title(' ( '+str(cnt[lmax[nc][0]])+' )')
t = arange(0.0, len(signal), 1)
ax2.axis([0, len(signal), minaxis, maxaxis])
ax2.plot(t,signal)
plt.axhline(linewidth=1, color='r', y=0)
fig.savefig(datainfo.dpath+'/Results/' + datainfo.name + '-' + s + ext + '-histo-sort.pdf', orientation='landscape', format='pdf')
# plt.show()
print '*******************'
for nc in range(nclusters):
lsignals.append((km.cluster_centers_[lmax[nc][0]], str(nc)+' ( '+str(cnt[lmax[nc][0]])+' )'))
#print s, np.max(km.cluster_centers_), np.min(km.cluster_centers_)
if nclusters % 2 == 0:
part = nclusters /2
else:
part = (nclusters /2) + 1
plotSignals(lsignals,part,2,np.max(km.cluster_centers_),np.min(km.cluster_centers_), datainfo.name + '-' + s + ext,
datainfo.name + '-' + s + ext, datainfo.dpath+'/Results/')
# for cc in range(nclusters):
# show_signal(km.cluster_centers_[cc])
f.close()
c = Counter(l)
print c
lcenters = []
numex = np.zeros(nc)
for i in c:
centers[i] /= c[i]
print i, c[i]
numex[i] = c[i]
lcenters.append((centers[i], 'center %d' % i))
mx = 0.26 # np.max(centers)
mn = -0.06 # np.min(centers)
plotSignals(lcenters, nc, 1, mx, mn, 'cluster-' + alg + '-%s-NC%d' % (line, nc),
'cluster-' + alg + '-%sNC%d' % (line, nc), clusterpath)
if alg == 'spectral':
params = {'clalg': 'spectral',
'nc': nc,
'labels': 'discretize',
'affinity': 'nearest_neighbors',
'n_neigbors': 30
}
elif alg == 'kmeans':
params = {'clalg': 'kmeans', 'nc': nc}
peakdata = {'labels': lab,
'centers': centers,
'numex': numex,
'params': params
}
numex = np.zeros(nc)
for i in c:
centers[i] /= c[i]
print i, c[i]
numex[i] = c[i]
lcenters.append((centers[i], 'center %d' % i))
cstd = np.zeros((nc, data.shape[1]))
for i in c:
center_mask = lab == i
cstd[i] = np.std(data[center_mask], axis=0)
mx = np.max(centers) # 0.26
mn = np.min(centers) #-0.06
plotSignals(lcenters, 8, 2, mx, mn, 'cluster-' + alg + '-N-%s-NC%d' % (line, nc),
'cluster-' + alg + '-%sNC%d' % (line, nc), clusterpath, cstd=cstd)
#
# if alg == 'spectral':
# params = {'clalg': 'spectral',
# 'nc': nc,
# 'labels': 'discretize',
# 'affinity': 'nearest_neighbors',
# 'n_neigbors': 30
# }
# elif alg == 'kmeans':
# params = {'clalg': 'kmeans', 'nc': nc}
# peakdata = {'labels': lab,
# 'centers': centers,
# 'numex': numex,
# 'params': params
maxaxis = np.max(centroids)
for nc in range(nclusters):
ax2 = fig.add_subplot(2, nclusters, nc + nclusters + 1)
signal = centroids[nc]
plt.title(' ( ' + str(nc + 1) + ' )')
t = np.arange(0.0, len(signal), 1)
ax2.axis([0, len(signal), minaxis, maxaxis])
ax2.plot(t, signal)
plt.axhline(linewidth=1, color='r', y=0)
fig.savefig(datainfo.dpath + '/' + datainfo.name + '/Results/' +
datainfo.name + '-' + sensor + '-' + str(nclusters) +
'-' + ext + '-histo-sort.pdf',
orientation='landscape',
format='pdf')
# plt.show()
print('*******************')
for nc in range(nclusters):
lsignals.append((centroids[nc], str(nc)))
if nclusters % 2 == 0:
part = nclusters / 2
else:
part = (nclusters / 2) + 1
plotSignals(
lsignals, part, 2, maxaxis, minaxis, datainfo.name + '-' +
sensor + '-' + str(nclusters) + '-' + ext,
datainfo.name + '-' + sensor + '-' + ext,
datainfo.dpath + '/' + datainfo.name + '/Results/')
datainfo.close_experiment_data(f)
ax.set_xticklabels(['class %d'% (i+1) for i in ind])
for i, h in enumerate(lhisto):
rects = ax.bar(ind+(i*width), h, width, color=colors[i])
fig.suptitle(datainfo.name + '-' + sensor+ '-' + ext, fontsize=48)
minaxis = np.min(centroids)
maxaxis = np.max(centroids)
for nc in range(nclusters):
ax2 = fig.add_subplot(2, nclusters, nc+nclusters+1)
signal = centroids[nc]
plt.title(' ( '+str(nc+1)+' )')
t = arange(0.0, len(signal), 1)
ax2.axis([0, len(signal), minaxis, maxaxis])
ax2.plot(t,signal)
plt.axhline(linewidth=1, color='r', y=0)
fig.savefig(datainfo.dpath + '/' + datainfo.name + '/Results/' + datainfo.name + '-' + sensor + '-' +
str(nclusters) + '-' + ext + '-histo-sort.pdf', orientation='landscape', format='pdf')
# plt.show()
print('*******************')
for nc in range(nclusters):
lsignals.append((centroids[nc], str(nc)))
if nclusters % 2 == 0:
part = nclusters /2
else:
part = (nclusters /2) + 1
plotSignals(lsignals, part, 2, maxaxis, minaxis, datainfo.name + '-' + sensor + '-' +
str(nclusters)+ '-' + ext, datainfo.name + '-' + sensor+ '-' + ext, datainfo.dpath + '/' + datainfo.name + '/Results/')
datainfo.close_experiment_data(f)
datainfo = experiments[expname]
f = h5py.File(datainfo.dpath + datainfo.name + ext + '.hdf5', 'r')
rslt = 50000
step = 10000
for dfile in range(0,len(datainfo.datafiles)):
d = f[datainfo.datafiles[dfile] + '/' + 'Raw']
length = int(d.shape[0]/float(step))
lsignalsvar = []
lsignalsmean = []
for s in [4,5,6,7]: # range(len(datainfo.sensors)):
print dfile, datainfo.sensors[s]
pvar = np.zeros(length)
pmean = np.zeros(length)
for pos in range(length):
pvar[pos] = np.std(d[pos*step:(pos*step)+rslt, s])
#print p
lsignalsvar.append((pvar,datainfo.sensors[s]))
plt.subplots(figsize=(20, 10))
plt.axis([0, length, 0, 0.2])
plot(range(length), pvar)
#plt.show()
plt.title(datainfo.datafiles[dfile]+ '-' + datainfo.sensors[s], fontsize=48)
plt.savefig(datainfo.dpath + '/Results/' + datainfo.datafiles[dfile] + '-' + datainfo.sensors[s]
+ '-variance.pdf', orientation='landscape', format='pdf')
plt.close()
plotSignals(lsignalsvar, 2, 2, 0.2, 0,
datainfo.datafiles[dfile]+'-'+str(rslt)+'-'+str(step)+'-variance',
datainfo.datafiles[dfile],
datainfo.dpath + '/Results/', orientation='portrait', cstd=[0.05, 0.05, 0.05, 0.05])
datainfo = experiments[expname]
f = h5py.File(datainfo.dpath + datainfo.name + '/' + datainfo.name + '.hdf5', 'r')
rslt = 50000
step = 10000
for dfile in range(0,len(datainfo.datafiles)):
d = f[datainfo.datafiles[dfile] + '/' + 'Raw']
length = int(d.shape[0]/float(step))
lsignalsvar = []
lsignalsmean = []
for s in range(len(datainfo.sensors)):
print dfile, datainfo.sensors[s]
pvar = np.zeros(length)
pmean = np.zeros(length)
for pos in range(length):
pvar[pos] = np.std(d[pos*step:(pos*step)+rslt, s])
#print p
lsignalsvar.append((pvar,datainfo.sensors[s]))
plt.subplots(figsize=(20, 10))
plt.axis([0, length, 0, 0.2])
plot(range(length), pvar)
#plt.show()
plt.title(datainfo.datafiles[dfile]+'-'+ datainfo.expnames[dfile]+ '-' + datainfo.sensors[s], fontsize=48)
plt.savefig(datainfo.dpath + '/' + datainfo.name + '/Results/' + 'variance-' + datainfo.datafiles[dfile]+'-'+ datainfo.expnames[dfile] + '-' + datainfo.sensors[s]
+ '.pdf', orientation='landscape', format='pdf')
plt.close()
plotSignals(lsignalsvar, 6, 2, 0.2, 0,
'variance-'+datainfo.datafiles[dfile]+'-'+datainfo.expnames[dfile]+'-'+str(rslt)+'-'+str(step),
datainfo.datafiles[dfile]+'-'+datainfo.expnames[dfile],
datainfo.dpath + '/' + datainfo.name + '/Results/', orientation='portrait', cstd=[0.05]*12)
ax.set_xticklabels(ind)
for i, h in enumerate(lhisto):
rects = ax.bar(ind+(i*width), h, width, color=colors[i])
fig.suptitle(datainfo.name + '-' + sensor, fontsize=48)
minaxis = np.min(centroids)
maxaxis = np.max(centroids)
for nc in range(nclusters):
ax2 = fig.add_subplot(2, nclusters, nc+nclusters+1)
signal = centroids[lmax[nc][0]]
plt.title(' ( '+str(cnt[lmax[nc][0]])+' )')
t = arange(0.0, len(signal), 1)
ax2.axis([0, len(signal), minaxis, maxaxis])
ax2.plot(t,signal)
plt.axhline(linewidth=1, color='r', y=0)
fig.savefig(datainfo.dpath + '/' + datainfo.name + '/Results/' + datainfo.name + '-' + sensor + '-' + str(nclusters)
+ '-histo-sort.pdf', orientation='landscape', format='pdf')
# plt.show()
print('*******************')
for nc in range(nclusters):
lsignals.append((centroids[lmax[nc][0]], str(nc)+' ( '+str(cnt[lmax[nc][0]])+' )'))
if nclusters % 2 == 0:
part = nclusters /2
else:
part = (nclusters /2) + 1
plotSignals(lsignals, part, 2, maxaxis, minaxis, datainfo.name + '-' + sensor,
datainfo.name + '-' + sensor, datainfo.dpath + '/' + datainfo.name + '/Results/')
datainfo.close_experiment_data(f)