def get_prediction(): loaded_model = joblib.load('data/decision_tree/model.pkl') date_string = request.args.get('date') date = datetime.strptime(date_string, '%Y-%m-%d') product = products[request.args.get("item_nbr")] data = { "date": date_string, "item_nbr": request.args.get("item_nbr"), "family": product['family'], "class": product['class'], "perishable": product['perishable'], "transactions": 1000, "year": date.year, "month": date.month, "day": date.day, "dayofweek": date.weekday(), "days_til_end_of_data": 0, "dayoff": date.weekday() >= 5 } df = pd.DataFrame(data=data, index=['row1']) df = decision_tree.encode_categorical_columns(df) pred = loaded_model.predict(df)t if FLUENTD_HOST: logger = sender.FluentSender(TENANT, host=FLUENTD_HOST, port=int(FLUENTD_PORT)) log_payload = {'prediction': pred[0], **data} print('logging {}'.format(log_payload)) if not logger.emit('prediction', log_payload): print(logger.last_error) logger.clear_last_error() return "%d" % pred[0]
def get_prediction(): loaded_model = joblib.load('data/decision_tree/model9.pkl') date_string = request.args.get('date') date = datetime.strptime(date_string, '%Y-%m-%d') data = { "date": date_string, "item_nbr": request.args.get("item_nbr"), "family": request.args.get("family"), "class": request.args.get("class"), "perishable": request.args.get("perishable"), "transactions": request.args.get("transactions"), "year": date.year, "month": date.month, "day": date.day, "dayofweek": date.weekday(), "days_til_end_of_data": 0, "dayoff": request.args.get("day_off") } df = pd.DataFrame(data=data, index=['row1']) df = decision_tree.encode_categorical_columns(df) pred = loaded_model.predict(df) return "%d" % pred[0]
def get_prediction(): #loaded_model = joblib.load('data/decision_tree/model.pkl') # mlflow.set_tracking_uri("http://35.247.183.209:5000") # mlflow.create_experiment("sales_prediction") tracking_uri = os.getenv('TRACKING_URI', "http://localhost:5000") experiment_name = os.getenv('EXPERIMENT_NAME', "global_experiments") mlflow.set_tracking_uri(tracking_uri) mlflow.set_experiment(experiment_name) artifact = mlflow.get_artifact_uri print("Experiment URI " + artifact) loaded_model = mlflow.sklearn.load_model( "model", "937b7254b4834a72b085bc55cdfbf460") date_string = request.args.get('date') date = datetime.strptime(date_string, '%Y-%m-%d') data = { "date": date_string, "item_nbr": request.args.get("item_nbr"), "family": request.args.get("family"), "class": request.args.get("class"), "perishable": request.args.get("perishable"), "transactions": request.args.get("transactions"), "year": date.year, "month": date.month, "day": date.day, "dayofweek": date.weekday(), "days_til_end_of_data": 0, "dayoff": request.args.get("day_off") } df = pd.DataFrame(data=data, index=['row1']) df = decision_tree.encode_categorical_columns(df) pred = loaded_model.predict(df) return "%d (From the model version : %s)" % (pred[0], artifact)