def test_verbose_int(verbose):
expected_line_count = {5: 2, False: 0, True: 10}
pool = Pool(TRAIN_FILE, column_description=CD_FILE)
tmpfile = 'test_data_dumps'
with LogStdout(open(tmpfile, 'w')):
cv(pool, {
"iterations": 10,
"random_seed": 0,
"loss_function": "Logloss"
},
verbose=verbose)
with open(tmpfile, 'r') as output:
assert (sum(1 for line in output) == expected_line_count[verbose])
with LogStdout(open(tmpfile, 'w')):
train(pool, {
"iterations": 10,
"random_seed": 0,
"loss_function": "Logloss"
},
verbose=verbose)
with open(tmpfile, 'r') as output:
assert (sum(1 for line in output) == expected_line_count[verbose])
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_cv_logging():
pool = Pool(TRAIN_FILE, column_description=CD_FILE)
cv(
pool, {
"iterations": 5,
"random_seed": 0,
"loss_function": "Logloss",
"json_log": JSON_LOG_PATH
})
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_cv_with_not_binarized_target():
train_file = data_file('adult_not_binarized', 'train_small')
cd = data_file('adult_not_binarized', 'train.cd')
pool = Pool(train_file, column_description=cd)
cv(
pool, {
"iterations": 5,
"random_seed": 0,
"loss_function": "Logloss",
"json_log": JSON_LOG_PATH
})
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_verbose_int(verbose):
expected_line_count = {5: 3, False: 0, True: 10}
pool = Pool(TRAIN_FILE, column_description=CD_FILE)
tmpfile = 'test_data_dumps'
with LogStdout(open(tmpfile, 'w')):
cv(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss"}, verbose=verbose)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == expected_line_count[verbose])
with LogStdout(open(tmpfile, 'w')):
train(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss"}, verbose=verbose)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == expected_line_count[verbose])
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_eval_set():
dataset = [(1, 2, 3, 4), (2, 2, 3, 4), (3, 2, 3, 4), (4, 2, 3, 4)]
labels = [1, 2, 3, 4]
train_pool = Pool(dataset, labels, cat_features=[0, 3, 2])
model = CatBoost({'learning_rate': 1, 'loss_function': 'RMSE', 'iterations': 2, 'random_seed': 0})
eval_dataset = [(5, 6, 6, 6), (6, 6, 6, 6)]
eval_labels = [5, 6]
eval_pool = (eval_dataset, eval_labels)
model.fit(train_pool, eval_set=eval_pool)
eval_pools = [eval_pool]
model.fit(train_pool, eval_set=eval_pools)
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_eval_set():
dataset = [(1, 2, 3, 4), (2, 2, 3, 4), (3, 2, 3, 4), (4, 2, 3, 4)]
labels = [1, 2, 3, 4]
train_pool = Pool(dataset, labels, cat_features=[0, 3, 2])
model = CatBoost({'learning_rate': 1, 'loss_function': 'RMSE', 'iterations': 2, 'random_seed': 0, "json_log": JSON_LOG_PATH})
eval_dataset = [(5, 6, 6, 6), (6, 6, 6, 6)]
eval_labels = [5, 6]
eval_pool = (eval_dataset, eval_labels)
model.fit(train_pool, eval_set=eval_pool)
eval_pools = [eval_pool]
model.fit(train_pool, eval_set=eval_pools)
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_verbose_int():
pool = Pool(TRAIN_FILE, column_description=CD_FILE)
tmpfile = 'test_data_dumps'
with LogStdout(open(tmpfile, 'w')):
cv(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss"}, verbose=5)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 2)
with LogStdout(open(tmpfile, 'w')):
cv(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss"}, verbose=False)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 0)
with LogStdout(open(tmpfile, 'w')):
cv(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss"}, verbose=True)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 10)
log_files = []
for i in range(3):
log_files.append(JSON_LOG_PATH[:-5]+str(i)+JSON_LOG_PATH[-5:])
with LogStdout(open(tmpfile, 'w')):
train(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss", "json_log": log_files[0]}, verbose=5)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 2)
with LogStdout(open(tmpfile, 'w')):
train(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss", "json_log": log_files[1]}, verbose=False)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 0)
with LogStdout(open(tmpfile, 'w')):
train(pool, {"iterations": 10, "random_seed": 0, "loss_function": "Logloss", "json_log": log_files[2]}, verbose=True)
with open(tmpfile, 'r') as output:
assert(sum(1 for line in output) == 10)
canonical_files = []
for log_file in log_files:
canonical_files.append(local_canonical_file(remove_time_from_json(log_file)))
return canonical_files
def test_cv_with_not_binarized_target():
train_file = data_file('adult_not_binarized', 'train_small')
cd = data_file('adult_not_binarized', 'train.cd')
pool = Pool(train_file, column_description=cd)
cv(pool, {"iterations": 5, "random_seed": 0, "loss_function": "Logloss"})
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))
def test_cv_logging():
pool = Pool(TRAIN_FILE, column_description=CD_FILE)
cv(pool, {"iterations": 5, "random_seed": 0, "loss_function": "Logloss"})
return local_canonical_file(remove_time_from_json(JSON_LOG_PATH))