def test_edge1(self):
"""
Edge test on get_one_hot to check that ValueError is raised when label
in data is not present in list.
"""
with self.assertRaises(ValueError):
biome.get_one_hot(['cat', 'dog'], np.array([['kitten']]))
def test_shot1(self):
"""
One-shot test on get_one_hot to check that one-hot encoding is
correctly performed.
"""
self.assertTrue(
(biome.get_one_hot(['cat', 'dog'],
np.array([['cat'],
['dog']])) == np.array([[1, 0],
[0,
1]])).all())
import numpy as np
from sklearn.datasets import make_classification
import biome
x, y = make_classification()
x = np.array(x)
y = np.array(y)
y_data = []
for i in range(len(y)):
if y[i] == 1:
y_data.append(['cat'])
else:
y_data.append(['dog'])
y_np = np.array(y_data)
y_onehot = biome.get_one_hot(['cat', 'dog'], y_np)
class TestSelectModel(unittest.TestCase):
"""
This class contains 3 smokes tests and an edge test for the functions:
get_trained_models, evaluate_rank_models, and get_prediction. Since these
algorithms each have their own unittest, the accuracy of these models will
not be focused on.
"""
def test_smoke1(self):
"""
Smoke test on get_trained_models to check that a list is
returned and is the correct length given the input.
"""
model_list = ['mlp1', 'mlp3', 'dtree']
OTU_path = input('Please enter the relative path to the OTU data: ')
print('')
metadata_path = input(
'Please enter the relative path to the categorical data: ')
print('')
category_labels = list(
(input('Please list the categorical variables of interest: ')).split(','))
print('')
print('What models would you like to test?')
model_list = list(
input('See README.md for abbreviations. Type all if all models should be '
'tested: ').split(','))
print('')
x_data, y_data = biome.data_loader(OTU_path, metadata_path)
y_output = biome.get_one_hot(category_labels, y_data)
x_train_unscale, x_test_unscale, y_train, y_test =\
biome.split_train_test(x_data, y_output)
scaler = preprocessing.StandardScaler().fit(x_train_unscale)
x_train = scaler.transform(x_train_unscale)
x_test = scaler.transform(x_test_unscale)
model_out = biome.evaluate_rank_models(x_train, y_train, x_test, y_test,
model_list)
model_name = model_out[0]
best_model = model_out[1]
yes_no = input('Would you like to make a prediction with the best model? ')
print('')
while yes_no == 'yes' or yes_no == 'Yes':