def _boundary_sub(disc_clsdict,
corpus,
names,
label,
verbose,
n_jobs,
threshold=0.03):
eb = eval_from_bounds
if verbose:
print ' boundary ({2}): subsampled {0} files in {1} sets'\
.format(sum(map(len, names)), len(names), label)
with verb_print(' boundary ({0}): calculating scores'.format(label),
verbose, True, True, True):
disc_bounds = [
Boundaries(disc_clsdict.restrict(ns), threshold=threshold)
for ns in names
]
gold_bounds = [
Boundaries(corpus.restrict(ns), threshold=threshold)
for ns in names
]
with verb_print(' boundary ({0}): calculating scores'.format(label),
verbose, False, True, False):
p, r = izip(*Parallel(n_jobs=n_jobs, verbose=5 if verbose else 0,
pre_dispatch='2*n_jobs') \
(delayed(eb)(disc, gold)
for disc, gold in zip(disc_bounds, gold_bounds)))
p, r = np.fromiter(p, dtype=np.double), np.fromiter(r, dtype=np.double)
p, r = praggregate(p, r)
return p, r
def _boundary_sub(disc_clsdict, corpus, names, label, verbose, n_jobs):
eb = eval_from_bounds
if verbose:
print ' boundary ({2}): subsampled {0} files in {1} sets'\
.format(sum(map(len, names)), len(names), label)
with verb_print(' boundary ({0}): calculating scores'.format(label),
verbose, True, True, True):
disc_bounds = [Boundaries(disc_clsdict.restrict(ns)) for ns in names]
gold_bounds = [Boundaries(corpus.restrict(ns)) for ns in names]
#print(len(disc_bounds[0].bounds))
acc = 0
'''for element in disc_bounds:
for key in element.bounds.keys():
acc += len(element.bounds[key])
print(acc, len(element.bounds))'''
acc = 0
'''for element in gold_bounds:
for key in element.bounds.keys():
acc += len(element.bounds[key])
print(acc, len(element.bounds))
acc = 0'''
#print(element.bounds.keys()[:10], len(element.bounds))
#print(element.bounds[element.bounds.keys()[0]])
with verb_print(' boundary ({0}): calculating scores'.format(label),
verbose, False, True, False):
p, r = izip(*Parallel(n_jobs=n_jobs, verbose=5 if verbose else 0,
pre_dispatch='2*n_jobs') \
(delayed(eb)(disc, gold)
for disc, gold in zip(disc_bounds, gold_bounds)))
p, r = np.fromiter(p, dtype=np.double), np.fromiter(r, dtype=np.double)
p, r = praggregate(p, r)
return p, r