def run(self):
custom_out('Predict Node initiated')
model_str = model_in(sys.argv[4])
if model_str == 'xgbtolr':
#pass off functional call to helper files
#predictions = xgb_lr()
pass
elif not (model_str in function_mappings().keys()):
custom_out('Model input is not in available models')
else:
prices_model = pd.read_pickle(Train().output().path)
test_df = pd.read_csv(TestDataPreProcessing().output().path)
test_x = test_postprocess(model_str, test_df)
predictions = prices_model.predict(test_x)
model_str2 = model_in(sys.argv[5])
if not (model_str2 in function_mappings().keys()):
custom_out('No (or invalid) 2nd model choice')
else:
second_model = pd.read_pickle(WeightedModel().output().path)
test_x2 = test_postprocess(model_str2, test_df)
predictions2 = second_model.predict(test_x2)
predictions = stacked_model_compute(predictions, predictions2)
submission = prediction_to_submission(test_df, predictions, model_str)
submission.to_csv(self.output().path, index=False)
custom_out('Write of submission to csv successful')
def run(self):
custom_out('Train Node initiated')
train_df = pd.read_csv(self.input().path)
model_str = model_in(sys.argv[4])
custom_out('Model chosen: {}'.format(model_str))
train_df = preprocess_data(train_df, model_str, 'train')
prices_model = model_choice(model_str, train_df)
custom_out('Successfully trained model')
with open(self.output().path, 'wb') as f:
pickle.dump(prices_model, f)
custom_out('Successfully wrote model to pickle')
def run(self):
model_str = model_in(sys.argv[5])
if model_str in function_mappings():
custom_out('Stacked Model Node initiated')
train_df = pd.read_csv(self.input().path)
custom_out('2nd Model chosen: {}'.format(model_str))
train_df = preprocess_data(train_df, model_str, 'train')
prices_model = model_choice(model_str, train_df)
custom_out('Successfully trained model')
with open(self.output().path, 'wb') as f:
pickle.dump(prices_model, f)
custom_out('Successfully wrote model to pickle')
else:
with open(self.output().path, 'wb') as f:
pickle.dump('none', f)
def run(self):
model_str = model_in(sys.argv[4])
print '\n\ndebug successful\n\n'