def test_repr(self):
m = ff.MeanZero()
k = ff.SqExpKernel()
l = ff.LikGauss()
regression_model = ff.GPModel(mean=m, kernel=k, likelihood=l)
print regression_model
print ff.repr_to_model(regression_model.__repr__())
assert regression_model == ff.repr_to_model(regression_model.__repr__())
def test_repr(self):
m = ff.MeanZero()
k = ff.SqExpKernel()
l = ff.LikGauss()
regression_model = ff.GPModel(mean=m, kernel=k, likelihood=l)
print regression_model
print ff.repr_to_model(regression_model.__repr__())
assert regression_model == ff.repr_to_model(
regression_model.__repr__())
def parse_results(results_filenames, max_level=None):
'''
Returns the best kernel in an experiment output file as a ScoredKernel
'''
if not isinstance(results_filenames, list):
# Backward compatibility wth specifying a single file
results_filenames = [results_filenames]
# Read relevant lines of file(s)
result_tuples = []
for results_filename in results_filenames:
lines = []
with open(results_filename) as results_file:
score = None
for line in results_file:
if line.startswith('score = '):
score = line[8:-2]
elif line.startswith("GPModel"):
lines.append(line)
elif (not max_level is None) and (len(re.findall('Level [0-9]+', line)) > 0):
level = int(line.split(' ')[2])
if level > max_level:
break
result_tuples += [ff.repr_to_model(line.strip()) for line in lines]
if not score is None:
best_tuple = sorted(result_tuples, key=lambda a_model : GPModel.score(a_model, score))[0]
else:
best_tuple = sorted(result_tuples, key=GPModel.score)[0]
return best_tuple
def parse_results(results_filenames, max_level=None):
'''
Returns the best kernel in an experiment output file as a ScoredKernel
'''
if not isinstance(results_filenames, list):
# Backward compatibility wth specifying a single file
results_filenames = [results_filenames]
# Read relevant lines of file(s)
result_tuples = []
for results_filename in results_filenames:
lines = []
with open(results_filename) as results_file:
score = None
for line in results_file:
if line.startswith('score = '):
score = line[8:-2]
elif line.startswith("GPModel"):
lines.append(line)
elif (not max_level is None) and (len(
re.findall('Level [0-9]+', line)) > 0):
level = int(line.split(' ')[2])
if level > max_level:
break
result_tuples += [ff.repr_to_model(line.strip()) for line in lines]
if not score is None:
best_tuple = sorted(
result_tuples,
key=lambda a_model: GPModel.score(a_model, score))[0]
else:
best_tuple = sorted(result_tuples, key=GPModel.score)[0]
return best_tuple
