def test_data_files_keys(self, tmpdir):
file = """
data_files:
train: 'tweets-train.csv'
test: 'tweets-test.csv'
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(ValueError) as ex:
c.set_configuration_parameters()
assert ex == "Missing {'target', 'features'} keys under data_files from context file."
def test_data_files_key_transformers(self, tmpdir):
file = """
data_files:
train: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
test: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
target: 'target_name'
features: 'features_name'
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(TypeError) as ex:
c.set_configuration_parameters()
assert " from config file expected to be a list got " in str(ex.value)
def test_data_files_keys_target_empty(self, tmpdir):
file = """
data_files:
train: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
test: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
target:
features:
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(ValueError) as ex:
c.set_configuration_parameters()
assert 'target cannot be empty.' == str(ex.value)
def test_data_files_keys_train_test(self, tmpdir):
file = """
data_files:
train: ''
test:
target:
features:
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(ValueError) as ex:
c.set_configuration_parameters()
assert "train cannot be empty." == str(ex.value)
def test_data_files_keys_train_csv_format(self, tmpdir):
file = """
data_files:
train: 'tweets-train.notcsv'
test:
target:
features:
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(TypeError) as ex:
c.set_configuration_parameters()
assert "train data must be a csv file with .csv extension." == str(
ex.value)
def test_data_files_key_error_factory_object(self, tmpdir):
file = """
data_files:
train: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
test: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
target: 'target_name'
features: 'features_name'
transformers:
- StopWordsRemoval:
estimators:
- s:
vectorizer:
- CountVectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(KeyError) as ex:
c.set_configuration_parameters()
def test_data_files_keys_train_csv_format_file_no_exist(self, tmpdir):
file = """
data_files:
train: 'tweets-train.csv'
test:
target:
features:
transformers:
estimators:
vectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(FileNotFoundError) as ex:
c.set_configuration_parameters()
assert "[Errno 2] File tweets-train.csv does not exist: 'tweets-train.csv'" == str(
ex.value)
def test_data_files_key_transformers(self, tmpdir):
file = """
data_files:
train: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
test: '$PROJECT_PATHS$/../data/raw/tweets-train.csv'
target: 'target_name'
features: 'features_name'
transformers:
- StopWordsRemoval:
estimators:
- s
vectorizer:
- CountVectorizer:
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(TypeError) as ex:
c.set_configuration_parameters()
assert "Parameter object: s must be of type dict " \
"found <class 'str'> instead." == str(ex.value)
def test_config_params_missing_keys(self, tmpdir):
file = """
data_files:
train: 'tweets-train.csv'
test: 'tweets-test.csv'
features: 'fe1'
target: 'sent'
transformers:
- transformer1:
param1: 'value_trans_1_param_1'
param2: 'value_trans_1_param_2'
- transformer2:
param1: 'value_trans_2_param_1'
param2: 'value_trans_2_param_2'
"""
p = tmpdir.mkdir("test").join("conf_file.yml")
p.write(file)
c = Context('exp_name_param', conf_file=str(p))
with pytest.raises(ValueError) as ex:
c.set_configuration_parameters()
assert ex == "Missing {'estimators', 'vectorizer'} compulsory keys in configuration file."
from mlpipeline.experiment import Experiment, Data
from mlpipeline.context import Context
from mlpipeline.pipeline import Pipeline
import pandas as pd
if __name__ == '__main__':
print("Starting ml pipeline program.")
# AUTOMATIC APPROACH #######################################
config = Context(conf_file='./experiment_configs/experiment002.yml')
config = config.set_configuration_parameters()
experiment = Experiment(config)
experiment.run()
print(experiment.results.confusion_matrix)
print(experiment.results.accuracy_score)
#print(experiment.results.predictions)
print("SUCCESSFUL!!!!!")
# END AUTOMATIC APPROACH ####################################
# context params
"""print(config.exp_name)
print(config.conf_file)
print(config.train)
print(config.test)
print(config.target)
print(config.features)
print(config.vectorizer)
print(config.estimators)
print(config.transformers)
print(" ")