test_predicted = fp.L1_test(dataset, f_preprocess, f_predict_model,
class_weights)
# Save files
directory = '../submissions/'
if not os.path.exists(directory):
os.makedirs(directory)
# Submission
submission = pd.concat((pd.DataFrame(rows), pd.DataFrame(test_predicted)),
axis=1)
submission.columns = ['record_id'] + ['start', 'end'] + activity_names
submission.to_csv('{}{}_submission.csv'.format(directory, name_to_save),
index=False)
if validate:
# Train & Predict L2_train
valid_predicted, scores = fp.L1_train(dataset,
f_preprocess,
f_predict_model,
class_weights,
verbose=1)
# L2_data
valid_predicted = pd.DataFrame(valid_predicted)
valid_predicted.columns = activity_names
valid_predicted.to_csv('{}{}_valid.csv'.format(directory, name_to_save),
index=False)
np.savetxt('{}{}_score.csv'.format(directory, name_to_save),
scores,
delimiter=",")
return(test_predicted / 2)
# Set Parameters
name_to_save = 'pl_L1_ET_vA1'
random_state = 1
prepbd_params = {'imputer_strategy':None}
f_preprocess = partial(fd.batch_preprocess, params=prepbd_params)
f_predict_model = partial(predict_model, random_state=random_state, class_weights=None)
# Train & Predict L1_train
valid_predicted, scores = fp.L1_train(dataset, f_preprocess, f_predict_model)
# Train & Predict L1_test
test_predicted = fp.L1_test(dataset, f_preprocess, f_predict_model, class_weights)
# Save files
directory = '../predict_location/'
if not os.path.exists(directory):
os.makedirs(directory)
# Submission
room_names = json.load(open('../public_data/rooms.json', 'r'))
submission = pd.concat((pd.DataFrame(rows), pd.DataFrame(test_predicted)), axis=1)
submission.columns = ['record_id'] + ['start', 'end'] + room_names
submission.to_csv('{}{}_submission.csv'.format(directory, name_to_save), index=False)
