def stats(self, alpha=0.05, start=0, batches=100):
"""
Generate posterior statistics for node.
"""
from utils import hpd, quantiles
from numpy import sqrt
try:
trace = np.squeeze(np.array(self.trace(), float)[start:])
n = len(trace)
if not n:
print 'Cannot generate statistics for zero-length trace in', self.__name__
return
return {
'n': n,
'standard deviation': trace.std(0),
'mean': trace.mean(0),
'%s%s HPD interval' % (int(100*(1-alpha)),'%'): hpd(trace, alpha),
'mc error': batchsd(trace, batches),
'quantiles': quantiles(trace)
}
except:
print 'Could not generate output statistics for', self.__name__
return
def write_vals(self,saveTo=None):
""" Saves estimated parameters to csv file.
Input:
- m (Model)
- saveTo (str) - path and name of csv file where results should be
saved. Defaults to mod.saveTo+'-posteriorValues.csv'.
"""
m=self
if saveTo==None:
fname=m.saveTo+'-posteriorValues.csv'
else:
fname=saveTo
f=open(fname,'w')
f.write('\t'.join(['Parameter','mean','median','95% HPD','std'])+'\n')
for v in m.parameters:
trac=getattr(m,v+'s')
hpdi=ut.hpd(trac,0.95)
form='%.2f'
if v.startswith('p') or v.startswith('k') or v.startswith('e'):
form='%.2e'
f.write('\t'.join([v,form%trac.mean(),form%sap(trac,50),('['+form+', '+form+']')%hpdi,form%trac.std()])+'\n')
f.close()
if saveTo==None:
print "Saved posterior median and confidence intervals for each parameter, see "+ m.name+'-posterior_values.csv'
else:
print "Saved posterior median and confidence intervals for each parameter, see "+ saveTo
def histogram(data, name, nbins=None, datarange=(None, None), format='png', suffix='', path='./', rows=1, columns=1, num=1, last=True, fontmap = {1:10, 2:8, 3:6, 4:5, 5:4}, verbose=1):
# Internal histogram specification for handling nested arrays
try:
# Stand-alone plot or subplot?
standalone = rows==1 and columns==1 and num==1
if standalone:
if verbose>0:
print 'Generating histogram of', name
figure()
subplot(rows, columns, num)
#Specify number of bins (10 as default)
uniquevals = len(unique(data))
nbins = nbins or uniquevals*(uniquevals<=25) or int(4 + 1.5*log(len(data)))
# Generate histogram
hist(data.tolist(), nbins, histtype='stepfilled')
xlim(datarange)
# Plot options
title('\n\n %s hist'%name, x=0., y=1., ha='left', va='top', fontsize='medium')
ylabel("Frequency", fontsize='x-small')
# Plot vertical lines for median and 95% HPD interval
quant = calc_quantiles(data)
axvline(x=quant[50], linewidth=2, color='black')
for q in hpd(data, 0.05):
axvline(x=q, linewidth=2, color='grey', linestyle='dotted')
# Smaller tick labels
tlabels = gca().get_xticklabels()
setp(tlabels, 'fontsize', fontmap[rows])
tlabels = gca().get_yticklabels()
setp(tlabels, 'fontsize', fontmap[rows])
if standalone:
if not os.path.exists(path):
os.mkdir(path)
if not path.endswith('/'):
path += '/'
# Save to file
savefig("%s%s%s.%s" % (path, name, suffix, format))
#close()
except OverflowError:
print '... cannot generate histogram'
def plot_coefs(coefs, label, offset, errorbars=True, marker='o'):
"Plots the coefficients of a logistic regression analysis"
x = np.array((0, 1, 2, 3))
if errorbars:
y = [coef_series.mean() for coef_series in coefs]
yerr = [[], []]
for yy, coef_series in zip(y, coefs):
uplim, lolim = utils.hpd(coef_series)
yerr[0].append(yy - lolim)
yerr[1].append(uplim - yy)
plt.errorbar(x + offset, y, yerr, fmt='none', ecolor='black')
else:
y = coefs
plt.plot(x + offset, y, marker, label=label)
def plot_stay_probs(probs, condition, legend):
"Plots stay probabilities with error bars"
y = [prob.median() for prob in probs]
plt.title(condition)
plt.ylim(0, 1)
plt.bar((0, 2), (y[0], y[2]), color=COLOR_COMMON, label='Common')
plt.bar((1, 3), (y[1], y[3]), color=COLOR_RARE, label='Rare')
plt.xticks((0.5, 2.5), ('Rewarded', 'Unrewarded'))
yerr = [[], []]
for yy, prob_series in zip(y, probs):
uplim, lolim = hpd(prob_series)
yerr[0].append(yy - lolim)
yerr[1].append(uplim - yy)
plt.errorbar((0, 1, 2, 3), y, yerr, fmt='none', ecolor='black')
plt.xlabel('Previous outcome')
plt.ylabel('Stay probability')
if legend:
plt.legend(loc='best', title='Previous transition')
def stats(self, alpha=0.05, start=0, batches=100, chain=None):
"""
Generate posterior statistics for node.
:Parameters:
alpha : float
The alpha level for generating posterior intervals. Defaults to
0.05.
start : int
The starting index from which to summarize (each) chain. Defaults
to zero.
batches : int
Batch size for calculating standard deviation for non-independent
samples. Defaults to 100.
chain : int
The index for which chain to summarize. Defaults to None (all
chains).
"""
from utils import hpd, quantiles
from numpy import sqrt
try:
trace = np.squeeze(np.array(self.trace(burn=start, chain=chain), float))
n = len(trace)
if not n:
print 'Cannot generate statistics for zero-length trace in', self.__name__
return
return {
'n': n,
'standard deviation': trace.std(0),
'mean': trace.mean(0),
'%s%s HPD interval' % (int(100*(1-alpha)),'%'): hpd(trace, alpha),
'mc error': batchsd(trace, batches),
'quantiles': quantiles(trace)
}
except:
print 'Could not generate output statistics for', self.__name__
return
def interval(self, alpha):
return tuple(hpd(self, 1 - alpha))
