def post(self):
local_variable = copy.deepcopy(GLOBAL_VARIABLE)
if "model" in local_variable:
json_data = request.get_json(force=True)
data_frame = prepare_single_job(json_data)
unnormalized_data = data_frame.copy()
normalized_data =\
prepare_single_job_jobtype_model(unnormalized_data,
TARGET_NAME,
local_variable['columns'],
local_variable['min'],
local_variable['max'],
local_variable['vectorizers'])
# No min,max needed for denormalization, since target
# is categorical (Hourly/Fixed)
prediction = predict(normalized_data, TARGET_NAME,
local_variable['model'])
app.logger.info("{} Prediction: {}".format(TARGET_NAME,
prediction))
return {
TARGET_NAME: {
'prediction': prediction,
'stats': local_variable['feature_importances']
}
}
return {}
def post(self):
local_variable = copy.deepcopy(GLOBAL_VARIABLE)
if local_variable.has_key("clusters"):
json_data = request.get_json(force=True)
data_frame = prepare_single_job(json_data)
unnormalized_data = data_frame.copy()
normalized_data, target_columns = prepare_single_job_clustering(
data_frame,
local_variable["centroids"].columns,
local_variable["min"],
local_variable["max"],
vectorizers=local_variable["vectorizers"],
weighting=True)
unnormalized_data = predict(unnormalized_data, normalized_data,
local_variable["clusters"],
local_variable["centroids"],
target_columns)
predictions = unnormalized_data.to_dict('records')
print "\n\n### Predictions:"
print predictions
return predictions[0]
else:
return {}
def post(self):
local_variable = copy.deepcopy(GLOBAL_VARIABLE)
if "model" in local_variable:
json_data = request.get_json(force=True)
data_frame = prepare_single_job(json_data)
unnormalized_data = data_frame.copy()
if 'job_type' in unnormalized_data and \
unnormalized_data['job_type'][0] != 'Hourly':
normalized_data =\
prepare_single_job_budget_model(unnormalized_data,
TARGET_NAME,
local_variable['columns'],
local_variable['min'],
local_variable['max'],
local_variable['vectorizers'])
prediction = predict(normalized_data, TARGET_NAME,
local_variable['model'],
local_variable['min'],
local_variable['max'])
app.logger.info("{} Prediction: {}".format(
TARGET_NAME, prediction))
return {
TARGET_NAME: {
'prediction': prediction,
'stats': local_variable['feature_importances']
}
}
return {}
def get_knn_prediction(json_data, local_variable, target="text"):
if target == "text":
get_example_text = True
else:
get_example_text = False
data_frame = prepare_single_job(json_data)
unnormalized_data = data_frame.copy()
normalized_data, target_columns = prepare_single_job_clustering(
data_frame,
local_variable["data_frame_normalized"].columns,
local_variable["min"],
local_variable["max"],
vectorizers=local_variable["vectorizers"],
weighting=True,
do_log_transform=True)
if target == "budget":
target_columns.append(target)
# Reweighting harms budget! Benefits subcategory however and shows more similar example snippet and title
predictions = predict_knn(unnormalized_data,
local_variable["data_frame_original"],
normalized_data,
local_variable["data_frame_normalized"],
k=K,
target_columns=target_columns,
do_reweighting=get_example_text,
get_example_text=get_example_text)
predictions = predictions.to_dict('records')
print "\n\n### Predictions:"
print predictions
return predictions[0]
def post(self):
local_variable = copy.deepcopy(GLOBAL_VARIABLE)
if "model" in local_variable:
try:
json_data = request.get_json(force=True)
data_frame = prepare_single_job(json_data)
unnormalized_data = data_frame.copy()
normalized_data =\
prepare_single_job_feedback_model(unnormalized_data,
TARGET_NAME,
local_variable['columns'],
local_variable['min'],
local_variable['max'],
local_variable['vectorizers'])
prediction = predict(normalized_data, TARGET_NAME,
local_variable['model'],
local_variable['min'],
local_variable['max'])
app.logger.info("{} Prediction: {}".format(
TARGET_NAME, prediction))
return {
TARGET_NAME: {
'prediction': prediction,
'stats': local_variable['feature_importances']
}
}
except Exception as e:
app.logger.error(e)
return {}
return {}