def test_counting(self):
ballots = [
{'Awesome': 2},
{'Slice': 1, 'Awesome': 1},
{'Awesome': 2},
{'Slice': 2},
{'Bluesy': 1},
{}
]
winners = cv.run_cv(ballots, 1)
self.assertEqual(winners, ['Awesome'])
winners = cv.run_cv(ballots, 2)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice']))
winners = cv.run_cv(ballots, 3)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice', 'Bluesy']))
def test_radio_values(self):
ballots = [{
'Blueberry': 1,
'Strawberry': 0,
'Banana': 0
}, {
'Blueberry': 1,
'Strawberry': 0,
'Banana': 0
}, {
'Blueberry': 1,
'Strawberry': 0,
'Banana': 0
}, {
'Blueberry': 0,
'Strawberry': 1,
'Banana': 0
}, {
'Blueberry': 0,
'Strawberry': 1,
'Banana': 0
}, {
'Blueberry': 0,
'Strawberry': 1,
'Banana': 0
}]
winners = cv.run_cv(ballots, 1)
self.assertSetEqual(set(winners), set(['Strawberry', 'Blueberry']))
def test_counting(self):
ballots = [{
'Awesome': 2
}, {
'Slice': 1,
'Awesome': 1
}, {
'Awesome': 2
}, {
'Slice': 2
}, {
'Bluesy': 1
}, {}]
winners = cv.run_cv(ballots, 1)
self.assertEqual(winners, ['Awesome'])
winners = cv.run_cv(ballots, 2)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice']))
winners = cv.run_cv(ballots, 3)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice', 'Bluesy']))
def test_radio_values(self):
ballots = [
{'Blueberry': 1, 'Strawberry': 0, 'Banana': 0},
{'Blueberry': 1, 'Strawberry': 0, 'Banana': 0},
{'Blueberry': 1, 'Strawberry': 0, 'Banana': 0},
{'Blueberry': 0, 'Strawberry': 1, 'Banana': 0},
{'Blueberry': 0, 'Strawberry': 1, 'Banana': 0},
{'Blueberry': 0, 'Strawberry': 1, 'Banana': 0}
]
winners = cv.run_cv(ballots, 1)
self.assertSetEqual(set(winners), set(['Strawberry', 'Blueberry']))
if hasattr(flags, 'cpu') and flags.cpu:
os.environ["CUDA_VISIBLE_DEVICES"] = ""
elif hasattr(flags, 'gpu') and flags.gpu is not None:
os.environ["CUDA_VISIBLE_DEVICES"] = ','.join(
str(i) for i in flags.gpu)
if flags.command_name == 'train':
import training
training.run_train(flags)
elif flags.command_name == 'test':
import testing
testing.run_test(flags)
elif flags.command_name == 'cv':
import cv
cv.run_cv(flags)
elif flags.command_name == 'plot':
import plotting
plotting.run_plot(flags)
elif flags.command_name == 'new':
import util
util.run_new(flags)
elif flags.command_name == 'explore':
import explore
explore.run_explore(flags)
elif flags.command_name == 'tftest':
import tftesting
tftesting.run_tftest(flags)
else:
parser.print_help()
def test_simple_tie(self):
ballots = [
{'Awesome': 5, 'Slice': 5}
]
winners = cv.run_cv(ballots, 1)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice']))
def test_simple(self):
ballots = [
{'Awesome': 5}
]
winners = cv.run_cv(ballots, 1)
self.assertEqual(winners, ['Awesome'])
def test_empty(self):
for ballots in [[], [{}, {}]]:
winners = cv.run_cv(ballots, 2)
self.assertEqual(winners, [])
def test_simple_tie(self):
ballots = [{'Awesome': 5, 'Slice': 5}]
winners = cv.run_cv(ballots, 1)
self.assertSetEqual(set(winners), set(['Awesome', 'Slice']))
def test_simple(self):
ballots = [{'Awesome': 5}]
winners = cv.run_cv(ballots, 1)
self.assertEqual(winners, ['Awesome'])
def test_empty(self):
for ballots in [[], [{}, {}]]:
winners = cv.run_cv(ballots, 2)
self.assertEqual(winners, [])
def main():
if len(sys.argv) - 1 < NUM_REQ_ARGS:
print(REQ_MESSAGE)
return
model_name = str(sys.argv[1])
data_path = str(sys.argv[2])
pickle_path = str(sys.argv[3])
num_folds = None
if model_name not in AVAILABLE_MODELS:
print("Unrecognized model: ", model_name, "\nAvailable models are", ", ".join(AVAILABLE_MODELS))
return
model_class = getattr(sys.modules[__name__], model_name)
labels = model_class.supervised
if labels:
if len(sys.argv) - 1 < NUM_REQ_ARGS + 1:
print(REQ_MESSAGE)
return
num_folds = int(sys.argv[NUM_REQ_ARGS + 1])
hyperparam_df = model_class.hyperparam_df
hyperparam_df = get_hyperparam_selections(hyperparam_df)
data = shuffle(pandas.read_csv(data_path))
while True:
reducer = input("Select dimensionality reducer from the list: " + ", ".join(AVAILABLE_DIM_REDUCERS) + " or press enter to skip: ")
if reducer and str(reducer) in AVAILABLE_DIM_REDUCERS:
reducer_class = getattr(sys.modules[__name__], reducer)
reducer_hdf = get_hyperparam_selections(reducer_class.recuder_hdf)
model=model_class(hyperparam_df, reducer_class, reducer_hdf)
break
elif not reducer:
# they don't want to set a reducer
model=model_class(hyperparam_df, None, None)
break
data = model.encode_categorical(data)
if labels:
print("Using column '"+ str(data.columns[-1]) + "' as labels.\n")
data_y = data.iloc[:,-1].values.tolist()
data.drop(data.columns[[-1]], axis=1, inplace=True)
data_X = data.values.tolist()
if num_folds > 0:
cv.run_cv(data_X, num_folds, data_y, len(set(data_y)), model)
model.fit((data_X, data_y)) # the final model to be saved
else:
if data.columns[-1].lower() == 'class':
data.drop(data.columns[[-1]], axis=1, inplace=True)
data_X = data.values.tolist()
model.fit(data_X)
pickle.dump(model, open(pickle_path, "wb" ))
