def graph_rocchio(fname):
with open(fname) as f:
results = pickle.load(f)
for percent in results.keys():
x = []
y = []
z = []
for alpha in results[percent]:
for beta in results[percent][alpha]:
x.append(alpha)
y.append(beta)
recalls = results[percent][alpha][beta]['recalls']
z.append(sum(recalls) / len(recalls))
fig = plt.figure()
ax = fig.gca(projection='3d')
xi = np.linspace(min(x), max(x))
yi = np.linspace(min(y), max(y))
X, Y = np.meshgrid(xi, yi)
Z = griddata(x, y, z, xi, yi)
surf = ax.plot_surface(X, Y, Z, rstride=5, cstride=5, cmap=cm.jet,
linewidth=1, antialiased=True)
ax.set_zlim3d(np.min(Z), np.max(Z))
fig.colorbar(surf)
ax.set_xlabel('alpha')
ax.set_ylabel('beta')
ax.set_zlabel('recall')
plt.show()
def plot_signal(signal, shape=None, name=None):
"""
Plot a numeric signal as an image.
"""
s = np.copy(signal)
if s.max() != 0.:
s *= 255.0/s.max()
s = s.astype(int)
if shape:
s = s.reshape(shape)
plt.figure()
plt.imshow(s, cmap=plt.cm.gray, interpolation='nearest')
if name:
plt.savefig(name)
plt.close()
def show_timeorder_info(Dt, mesh_sizes, errors):
'''Performs consistency check for the given problem/method combination and
show some information about it. Useful for debugging.
'''
# Compute the numerical order of convergence.
orders = {}
for key in errors:
orders[key] = _compute_numerical_order_of_convergence(Dt, errors[key])
# Print the data to the screen
for i, mesh_size in enumerate(mesh_sizes):
print
print('Mesh size %d:' % mesh_size)
print('dt = %e' % Dt[0]),
for label, e in errors.items():
print(' err_%s = %e' % (label, e[i][0])),
print
for j in range(len(Dt) - 1):
print(' '),
for label, o in orders.items():
print(' ord_%s = %e' % (label, o[i][j])),
print
print('dt = %e' % Dt[j+1]),
for label, e in errors.items():
print(' err_%s = %e' % (label, e[i][j+1])),
print
# Create a figure
for label, err in errors.items():
pp.figure()
ax = pp.axes()
# Plot the actual data.
for i, mesh_size in enumerate(mesh_sizes):
pp.loglog(Dt, err[i], '-o', label=mesh_size)
# Compare with order curves.
pp.autoscale(False)
e0 = err[-1][0]
for o in range(7):
pp.loglog([Dt[0], Dt[-1]],
[e0, e0 * (Dt[-1] / Dt[0]) ** o],
color='0.7')
pp.xlabel('dt')
pp.ylabel('||%s-%s_h||' % (label, label))
# pp.title('Method: %s' % method['name'])
ppl.legend(ax, loc=4)
pp.show()
return
def new_axes(self, figsize):
fig = plt.figure(figsize=(figsize, figsize),
facecolor='w',
edgecolor='w')
rect = [0.1, 0.1, 0.8, 0.8]
ax = WindroseAxes(fig, rect, axisbg='w')
fig.add_axes(ax)
return fig, ax
def layerActivations(model, data, labels):
print('Visualizing activations with tSNE...')
if not os.path.exists(cc.cfg['plots']['layer_activations_dir']):
os.makedirs(cc.cfg['plots']['layer_activations_dir'])
numLabels = cc.cfg['plots']['layer_activations_label_cap']
data = data[:cc.cfg['plots']['layer_activations_points_cap']]
labels = labels[:numLabels,:cc.cfg['plots']['layer_activations_points_cap']]
subplotCols = numLabels
subplotRows = len(model.layers)-1
subplotIdx = 1
plt.figure(figsize=(5*subplotCols,5*subplotRows))
for i in range(1,len(model.layers)):
print('Running tSNE for layer {}/{}'.format(i+1,len(model.layers)))
func = K.function([model.layers[0].input], [model.layers[i].output])
out = func([data])[0]
tsneModel = TSNE(n_components = 2, random_state = 0)
tsneOut = tsneModel.fit_transform(out).T
# labeledTsneOut = np.hstack((tsneOut, labels[0].reshape(-1,1)))
for j in range(numLabels):
plt.subplot(subplotRows, subplotCols, subplotIdx)
plt.title('{} / {}'.format(model.layers[i].name,cc.exp['params']['data']['labels'][j]))
plt.scatter(tsneOut[0],tsneOut[1],c=labels[j],cmap = 'plasma')
subplotIdx += 1
# tsneDF = pd.DataFrame(labeledTsneOut, columns = ('a', 'b', 'c'))
# plot = tsneDF.plot.scatter(x = 'a', y = 'b', c = 'c', cmap = 'plasma')
plt.tight_layout()
plt.savefig('{}/activations.png'.format(cc.cfg['plots']['layer_activations_dir']))
plt.close()
print('...done')
def histograms(modelLogger):
if not cc.cfg['plots']['histograms']:
return
if not os.path.exists(cc.cfg['plots']['histograms_dir']):
os.makedirs(cc.cfg['plots']['histograms_dir'])
cntEpochs = len(modelLogger.epochLogs)
cntLayers = len(modelLogger.epochLogs[-1]['weights'])
logVals = [
{'name':'weights','color':'blue'},
{'name':'updates','color':'red'},
{'name':'ratios','color':'green'}
]
subplotRows = len(logVals)
subplotCols = cntLayers
for x in range(cntEpochs):
subplotIdx = 1
plt.figure(figsize=(5*subplotCols,5*subplotRows))
plt.suptitle('Histograms per layer, epoch {}/{}'.format(x,cntEpochs-1), fontsize=14)
for logVal in logVals:
for i,layer in enumerate(modelLogger.epochLogs[x][logVal['name']]):
histmin = layer.min()
histmax = layer.max()
plt.subplot(subplotRows, subplotCols, subplotIdx)
plt.title('{}, {}'.format(modelLogger.model.layers[modelLogger.loggedLayers[i]].name,logVal['name']))
plt.hist(layer, range = (histmin, histmax), bins = 30, color = logVal['color'])
subplotIdx+=1
plt.tight_layout()
plt.subplots_adjust(top=0.9)
plt.savefig('{}/hist_{e:03d}.png'.format(cc.cfg['plots']['histograms_dir'], e = x))
plt.close()
env.reset(np.array([0.0]))
code.reset()
[mp,varp,spsp,sp,msep,parts,ws] = pf.particle_filter(code, env, timewindow=timewindow,
dt=dt, nparticles=nparticles,
mode='v', testf=(lambda x:x))
if gaussian:
print "MSE of gaussian filter %f"% mseg
print "MSE of particle filter %f"% msep
if plotting:
#matplotlib.rcParams['font.size']=10
plt.close()
plt.figure()
ax1 = plt.gcf().add_subplot(2,1,1)
times = np.arange(0.0,dt*timewindow,dt)
if gaussian:
ax1.plot(times,sg,'r',label='Signal')
if sum(sum(spsg)) !=0:
(ts,neurs) = np.where(spsg == 1)
spiketimes = times[ts]
thetas = [code.neurons[i].theta for i in neurs]
else:
spiketimes = []
thetas = []
ax1.plot(spiketimes,thetas,'yo',label='Spike times')
ax1.plot(times,mg,'b',label='Mean prediction')
ax1.set_title('Gaussian Filter')
def new_axes(self, figsize):
fig = plt.figure(figsize=(figsize, figsize), facecolor='w', edgecolor='w')
rect = [0.1, 0.1, 0.8, 0.8]
ax = WindroseAxes(fig, rect, axisbg='w')
fig.add_axes(ax)
return fig, ax
#ax.line([[10, 10], [16.5, 16.5]])
plt.xlabel("SDSS u magnitude, mag")
plt.ylabel("GC u magnitude, mag")
plt.savefig('test/u_SDSS_vs_GC')
'''
limits = [[12, 18.5], [11.4, 16.5], [10, 16], [9.5, 16], [9.5, 16]]
for i, band in enumerate(bands):
lim_lo = limits[i][0]
lim_hi = limits[i][1]
x, m, b = makeDiagonalLine([lim_lo, lim_hi])
fig = plt.figure(figsize=(8, 8))
ax = plt.subplot(111)
c = ppl.scatter(ax, petroMags[:, i], mags[:, i], s=8, c='k', edgecolor='k')
ax.axis([lim_lo, lim_hi, lim_lo, lim_hi])
ax.errorbar(petroMags[:, i], mags[:, i], yerr=mag_err[:, i], mew=0, linestyle = "none", color="black")
plt.plot(x,m*x + b, c='k')
ax.xaxis.set_major_locator(majorLocator)
ax.xaxis.set_minor_locator(minorLocator)
ax.yaxis.set_major_locator(majorLocator)
ax.yaxis.set_minor_locator(minorLocator)
ax.minorticks_on()
# Change the labels back to black
ax.xaxis.label.set_color('black')
ax.yaxis.label.set_color('black')
# Change the axis title also back to black
[78, 376],
[79, 164],
[80, 194],
[81, 342],
[82, 314],
[83, 68],
[84, 54],
[85, 131],
[86, 52],
[87, 220],
[88, 145],
[89, 63],
[90, 608],
[91, 68],
[92, 155],
[93, 49],
[94, 108],
[95, 59],
[96, 67],
[97, 58],
[98, 42],
[99, 210],
[100, 174]])
plt.figure(figsize=(5,5))
plt.scatter(histdata.T[0],histdata.T[1])
plt.savefig('hist.png')
plt.close()
# import matplotlib.pyplot as plt
import prettyplotlib as ppl # makes nicer colors and generally better to look at graphs
from prettyplotlib import plt # This is "import matplotlib.pyplot as plt" from the prettyplotlib library
from prettyplotlib import mpl # This is "import matplotlib as mpl" from the prettyplotlib library
# change font to Open Sans (has some kerning issues, though)
# mpl.rcParams.update({'font.family':'Open Sans'})
# get name of file to process
inputFileName = sys.argv[1] # keyPresses_Heatmap.csv
# load csv file with EPOCHTIME;NOFPROCESSES
data = np.loadtxt(inputFileName, delimiter=";")
# data[:,1] = [x - notActuallyChromeTabs for x in data[:,1]]
#
print data[:,0], data[:,1]
fig = plt.figure(figsize=(14,8))
ax = fig.add_subplot(111)
# ppl.bar(ax, data[:,0], data[:,1])
ax.bar(data[:,0], data[:,1])
# ax.axis('tight')
plt.show()
# xy = [(x_int[i],y_int[i]) for i in range(0,len(x_int))]
# xy_sorted = np.array(sorted(xy, key=lambda yz: yz[1], reverse=True))
# ax.bar(xy_sorted[:,0], xy_sorted[:,1])
##
import sys
if __name__ == '__main__':
# Accumulate the counts fed into standard in (stdin)
counts = []
for line in sys.stdin:
topic, count = line.split()
# Shift the ones up by a tiny amount to allow them to be visible on the graph
counts.append(int(count) + 0.05)
print 'Total page view counts: {}'.format(len(counts))
# Display the hourly page view distribution on a log-log plot
# This matches our friendly and well understood Zipf distribution
fig = plt.figure(figsize=(9, 5))
ax = fig.add_subplot(1, 1, 1)
plt.plot(xrange(len(counts)), counts, linewidth=3, label='Pageviews per page')
#
ax.set_xscale('log')
ax.set_yscale('log')
#
plt.title('Log-log plot of hourly Wikipedia page view distribution')
plt.xlabel('Rank order')
plt.ylabel('Frequency')
plt.grid(color='black')
plt.legend()
#
plt.savefig('hourly_wikipedia_zipf.pdf')
plt.show(block=True)
import prettyplotlib as ppl
from prettyplotlib import plt
import numpy
T = 20.
dt = 0.001
plt.figure(figsize=(6,5))
def sample_and_plot(X0, A, H, dt, N, NSamples, ax):
Xs = numpy.zeros((NSamples,N,2))
Xs[:,0] = X0
for i in range(1,N):
for j in range(NSamples):
Xs[j,i] = Xs[j,i-1] +dt*numpy.dot(A,Xs[j,i-1])\
+ numpy.sqrt(dt)*numpy.dot(H, numpy.random.normal(size=(2,)))
ts = numpy.arange(0.0,N*dt,dt)
[ppl.plot(ts,Xs[i,:,0],ax=ax) for i in range(NSamples)]
def sample(X0, A, H, dt, N, NSamples):
Xs = numpy.zeros((NSamples,N,2))
Xs[:,0] = X0
for i in range(1,N):
for j in range(NSamples):
Xs[j,i] = Xs[j,i-1] +dt*numpy.dot(A,Xs[j,i-1])\
+ numpy.sqrt(dt)*numpy.dot(H, numpy.random.normal(size=(2,)))
ts = numpy.arange(0.0,N*dt,dt)
return ts, Xs
cax = axs[1].imshow(masked,interpolation='nearest',aspect='auto', cmap = cmap)
plt.colorbar(cax)
axs[0].hist(sanitized.ravel(),bins=100,color='k')
map(artist.adjust_spines,axs)
axs[0].set_ylabel(artist.format('Frequency'))
axs[0].set_xlabel(artist.format('Semantic Distance'))
axs[1].set_ylabel(artist.format('Tweet'))
axs[1].set_xlabel(artist.format('Tweet'))
plt.tight_layout()
plt.show()
'''
pca =PCA(n_components=2)
pca.fit(sanitized)
decomp = plt.figure()
panel = decomp.add_subplot(111)
panel.scatter(pca.components_[0,:]*np.sqrt(pca.explained_variance_ratio_[0]),
pca.components_[1,:]*np.sqrt(pca.explained_variance_ratio_[1]),s=20,c='k')
artist.adjust_spines(panel)
panel.set_xlabel(r'\Large $\mathbf{1^{st}}$ \textbf{Semantic Dimension}')
panel.set_ylabel(r'\Large $\mathbf{2^{nd}}$ \textbf{Semantic Dimension}')
plt.show()
'''
x = SV('')
x.frequencies(open('../data/alcohol.txt','rb').read().replace('\n',''))
'''
def visualizeSequentialOutput(model, layerIdx, df):
if not os.path.exists(cc.cfg['plots']['seq_output_dir']):
os.makedirs(cc.cfg['plots']['seq_output_dir'])
if cc.cfg['plots']['seq_output_seq_input_name'] == 'smiles':
input = data.formatSequentialInput(df)
elif cc.cfg['plots']['seq_output_seq_input_name'] == 'fasta':
input = data.formatFastaInput(df)
else:
raise 'visual err'
# model.layers[layerIdx].return_sequences = True
# model.compile(loss="mean_squared_error", optimizer="rmsprop")
cfg = model.get_config()[:4]
cfg = model.get_config()[:layerIdx+1]
del cfg[2]
layerIdx -= 1
# print cfg
cfg[layerIdx]['config']['return_sequences'] = True
seqModel = Sequential.from_config(cfg)
seqModel.set_weights(model.get_weights())
seqModel.layers[layerIdx].return_sequences = True
outputFunction = K.function([seqModel.layers[0].input],
[seqModel.layers[layerIdx].output])
output = outputFunction([input])[0]
'''
sns.set()
for i,smilesOutput in enumerate(output):
g = sns.clustermap(smilesOutput.T, col_cluster=False, method='single',metric='cosine')
g.savefig('{}/seq_output.png'.format(cc.cfg['plots']['seq_output_dir']))
'''
dropSet = Set(cc.cfg['plots']['seq_output_ignore_neurons'])
if cc.cfg['plots']['seq_output_select_neurons']:
arrMask = cc.cfg['plots']['seq_output_select_neurons']
else:
arrMask = list(range(output.shape[2]))
arrMask = np.array([x for x in arrMask if not x in dropSet])
fig = plt.figure(figsize=(input.shape[1] * 0.3,len(arrMask) * len(df) * 1.5))
for i,seqOutput in enumerate(output):
# print seqOutput.shape
# print seqOutput
selected = seqOutput.T[arrMask]
Z = sch.linkage(selected, method='single', metric='cosine')
leaves = sch.leaves_list(Z)
# leaves = range(len(selected))
reordered = selected[leaves]
ax = fig.add_subplot(len(df),1,i+1)
print 'foo'
ppl.pcolormesh(fig, ax, reordered,
xticklabels=list(df.values[i][0]),
yticklabels=arrMask[leaves],
vmin=-1,
vmax=1)
print 'foo'
print 'bar'
fig.savefig('{}/{}'.format(cc.cfg['plots']['seq_output_dir'],cc.cfg['plots']['seq_output_name']))
print 'bar'
