def evaluate(sample_file, **kwargs):
use_duration = True if kwargs['D'] else False
window = int(kwargs['w'])
pv('window', 'use_duration', stdout=True)
samples = None
models = None
if kwargs['M'] or kwargs['B']:
samples = {classification.normalize_name(t): [t] for t in sample_file}
models = extract_models(ZIPPED_MODELS)
if samples and models:
if kwargs['M']:
classification.evaluate(samples, models,
use_duration=use_duration,
window=window)
if kwargs['B']:
binary_samples = classification.binarize(samples)
binary_models = classification.binarize(models)
if binary_samples and binary_models:
classification.evaluate(binary_samples, binary_models,
use_duration=use_duration,
window=window)
def main():
args = docopt(__doc__)
use_duration = True if args["-D"] else False
window = int(args["-w"])
models = model.extract_models(model.ZIPPED_MODELS)
if args["-B"]:
models = classification.binarize(models)
for line in sys.stdin:
splitted = line.strip().split("\t")
if len(splitted) == 5:
mid_terminal, uid, bid, aid, histories = splitted
term = classification.terminal_name(mid_terminal)
paths = stream.decode(histories)
assert len(paths) == 1
decision = classification.query(term, paths[0], models, use_duration=use_duration, window=window)
if decision:
forward(
mid_terminal,
uid,
bid,
aid,
decision.decision,
decision.prob,
decision.path,
decision.probs,
decision.principal,
)
