def raw_data_to_training_days(raw_data_path, training_days_path, cutoff_times):
    #df_flight_history = flighthistory.get_df_flight_history_from_raw_format(os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"))
 
    print(os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"))

    days_flight_ids = flighthistory.process_flight_history_to_train_day_files(
        input_path = os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"),
        output_path = training_days_path,
        output_folder_name = "FlightHistory",
        output_file_name = "flighthistory.csv",
        cutoff_times = cutoff_times,
        start_hours_offset = -9)

    print("Flight History Events")
    utilities.split_file_based_on_times_filter_on_ids_streaming(
        os.path.join(raw_data_path, "FlightHistory", "flighthistoryevents.csv"),
        training_days_path,
        "FlightHistory",
        "flighthistoryevents.csv",
        "flight_history_id",
        days_flight_ids)
    
    print("ASDI Flight Plan")
    days_flight_plan_ids = utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiflightplan.csv"), training_days_path,
        "ASDI", "asdiflightplan.csv", "flighthistoryid", days_flight_ids, ids_to_track_column_name="asdiflightplanid")

    print("ASDI Position")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiposition.csv"), training_days_path,
        "ASDI", "asdiposition.csv", "flighthistoryid", days_flight_ids)

    print("ASDI Airway")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiairway.csv"), training_days_path,
        "ASDI", "asdiairway.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPFix")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpfix.csv"), training_days_path,
        "ASDI", "asdifpfix.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPCenter")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpcenter.csv"), training_days_path,
        "ASDI", "asdifpcenter.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPSector")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpsector.csv"), training_days_path,
        "ASDI", "asdifpsector.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPWaypoint")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpwaypoint.csv"), training_days_path,
        "ASDI", "asdifpwaypoint.csv", "asdiflightplanid", days_flight_plan_ids)

    for ct in cutoff_times:
        print ct
        day_output_path = os.path.join(training_days_path, utilities.get_day_str(ct, -9))
        day_beginning, day_end = utilities.get_day_boundaries(ct, -9)

        if not os.path.exists(day_output_path):
            os.makedirs(day_output_path)
        weather.process_one_day(raw_data_path, day_output_path, day_beginning, day_end, "train")
def raw_data_to_training_days(raw_data_path, training_days_path, cutoff_times):
    #df_flight_history = flighthistory.get_df_flight_history_from_raw_format(os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"))
 
    print(os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"))

    days_flight_ids = flighthistory.process_flight_history_to_train_day_files(
        input_path = os.path.join(raw_data_path, "FlightHistory", "flighthistory.csv"),
        output_path = training_days_path,
        output_folder_name = "FlightHistory",
        output_file_name = "flighthistory.csv",
        cutoff_times = cutoff_times,
        start_hours_offset = -9)

    print("Flight History Events")
    utilities.split_file_based_on_times_filter_on_ids_streaming(
        os.path.join(raw_data_path, "FlightHistory", "flighthistoryevents.csv"),
        training_days_path,
        "FlightHistory",
        "flighthistoryevents.csv",
        "flight_history_id",
        days_flight_ids)
    
    print("ASDI Flight Plan")
    days_flight_plan_ids = utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiflightplan.csv"), training_days_path,
        "ASDI", "asdiflightplan.csv", "flighthistoryid", days_flight_ids, ids_to_track_column_name="asdiflightplanid")

    print("ASDI Position")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiposition.csv"), training_days_path,
        "ASDI", "asdiposition.csv", "flighthistoryid", days_flight_ids)

    print("ASDI Airway")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdiairway.csv"), training_days_path,
        "ASDI", "asdiairway.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPFix")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpfix.csv"), training_days_path,
        "ASDI", "asdifpfix.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPCenter")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpcenter.csv"), training_days_path,
        "ASDI", "asdifpcenter.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPSector")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpsector.csv"), training_days_path,
        "ASDI", "asdifpsector.csv", "asdiflightplanid", days_flight_plan_ids)

    print("ASDI FPWaypoint")
    utilities.split_file_based_on_times_filter_on_ids_streaming(os.path.join(raw_data_path, "ASDI", "asdifpwaypoint.csv"), training_days_path,
        "ASDI", "asdifpwaypoint.csv", "asdiflightplanid", days_flight_plan_ids)

    for ct in cutoff_times:
        print ct
        day_output_path = os.path.join(output_path, utilities.get_day_str(ct, -9))
        day_beginning, day_end = utilities.get_day_boundaries(ct, -9)

        if not os.path.exists(day_output_path):
            os.makedirs(day_output_path)
        weather.process_one_day(raw_data_path, day_output_path, day_beginning, day_end, "train")
Exemplo n.º 3
0
def training_day_to_test_day(training_day_path, test_day_path, solution_path, cutoff_time): 
    flighthistory.write_flight_history_test_day_file(
        os.path.join(training_day_path, "FlightHistory", "flighthistory.csv"),
        os.path.join(utilities.get_output_subdirectory(test_day_path, "FlightHistory"), "flighthistory.csv"),
        cutoff_time)

    flighthistory.write_flight_history_test_day_and_solution_test_flights_only(
        os.path.join(training_day_path, "FlightHistory", "flighthistory.csv"),
        os.path.join(test_day_path, "test_flights.csv"),
        os.path.join(solution_path, utilities.get_day_str(cutoff_time) + "_solution.csv"),
        cutoff_time)

    utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "FlightHistory", "flighthistoryevents.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "FlightHistory"), "flighthistoryevents.csv"),
        "date_time_recorded",
        utilities.parse_datetime_format3,
        cutoff_time)

    utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "ASDI", "asdiposition.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiposition.csv"),
        "received",
        utilities.parse_datetime_format1,
        cutoff_time)

    flight_plan_ids = utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "ASDI", "asdiflightplan.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiflightplan.csv"),
        "updatetimeutc",
        utilities.parse_datetime_format2,
        cutoff_time,
        ids_to_track_column_name = "asdiflightplanid")

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdiairway.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiairway.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpfix.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpfix.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpcenter.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpcenter.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpsector.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpsector.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpwaypoint.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpwaypoint.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    day_beginning, day_end = utilities.get_day_boundaries(cutoff_time)

    weather.process_one_day(
        training_day_path, 
        test_day_path, 
        day_beginning, 
        cutoff_time, 
        "test", 
        cutoff_time = cutoff_time)
def training_day_to_test_day(training_day_path, test_day_path, solution_path, cutoff_time): 
    flighthistory.write_flight_history_test_day_file(
        os.path.join(training_day_path, "FlightHistory", "flighthistory.csv"),
        os.path.join(utilities.get_output_subdirectory(test_day_path, "FlightHistory"), "flighthistory.csv"),
        cutoff_time)

    flighthistory.write_flight_history_test_day_and_solution_test_flights_only(
        os.path.join(training_day_path, "FlightHistory", "flighthistory.csv"),
        os.path.join(test_day_path, "test_flights.csv"),
        os.path.join(solution_path, utilities.get_day_str(cutoff_time) + "_solution.csv"),
        cutoff_time)
    
    utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "FlightHistory", "flighthistoryevents.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "FlightHistory"), "flighthistoryevents.csv"),
        "date_time_recorded",
        utilities.parse_datetime_format3,
        cutoff_time)

    utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "ASDI", "asdiposition.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiposition.csv"),
        "received",
        utilities.parse_datetime_format1,
        cutoff_time)

    flight_plan_ids = utilities.filter_file_based_on_cutoff_time_streaming(os.path.join(training_day_path, "ASDI", "asdiflightplan.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiflightplan.csv"),
        "updatetimeutc",
        utilities.parse_datetime_format2,
        cutoff_time,
        ids_to_track_column_name = "asdiflightplanid")

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdiairway.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdiairway.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpfix.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpfix.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpcenter.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpcenter.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpsector.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpsector.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    utilities.filter_file_based_on_ids_streaming(os.path.join(training_day_path, "ASDI", "asdifpwaypoint.csv"), 
        os.path.join(utilities.get_output_subdirectory(test_day_path, "ASDI"), "asdifpwaypoint.csv"),
        "asdiflightplanid",
        flight_plan_ids)

    day_beginning, day_end = utilities.get_day_boundaries(cutoff_time)

    weather.process_one_day(
        training_day_path, 
        test_day_path, 
        day_beginning, 
        cutoff_time, 
        "test", 
        cutoff_time = cutoff_time)