Skip to content

Latest commit

 

History

History
397 lines (380 loc) · 13.6 KB

README.md

File metadata and controls

397 lines (380 loc) · 13.6 KB

How to run ?

  • Create a virtual environment

  • Activate virtual environment

    • Linux
    source venv/bin/activate
    • Windows
    venv\Scripts\activate.bat

  • Install requirements

    pip3 install -r requirements.txt

  • Make sure to migrate first

    python3 manage.py migrate

  • Generate/Load data or use the existing db.sqlite3 file.

  • Cancel the flights

    • Use the following command to cancel x% of total flight randomly
    python3 manage.py cancel --percent 0.005
    • For more controlled cancellation, open the database in DB Browser for SQLite and change the status of the flight to Cancelled in the flight_flight table.

  • Run reallocation

    • Use the following command to run the reallocation
    python3 manage.py allocate --save "result.csv" --config "settings.yml"
    • This accepts two optional arguments
      • --save : to save the result in a csv file, default is result.csv
      • --config : to specify the config file, default is settings.yml
    • For more information on the config file refer to the Buisness Rules section below.

  • Run Post analysis

    • Three file should be generated, result.csv for pnr-flight allocation, result.txt for summary and result.flight_stats.csv for per flight stats.
    • This contains default solutions.
    • Alternatively Use the post_analysis.ipynb notebook to analyse the results.
    • The second cell contains the file name of the result file, change it accordingly.
    • Run all!!

Data Ingestion

Data Mapping

This section describes how we mapped the given data to the database tables.

Assumptions:

  • The flight_number is unique for each flight.
  • Aircraft's are unique ( from given data mapped using tail number )
  • The term Flight is used to refer Inventory
  • Multiple class belongs to a single cabin
  • PNR are mapped to class
    • Default cabin to class mapping have been defined
  • SSR, value added service etc are mapped to passengers and pnr scores are calculated on the time of ingestion

How to run ?

  • The settings.yml file contains some necessary information for data ingestion. Please strictly follow the format and naming conventions.

    • For more information on the config file refer to the Buisness Rules section below.

  • Run the flight/management/data/generate_basic.ipynb to generate some p reprocessed csv files.

  • Run the following command to finally ingest data.

python3 manage.py flush --noinput && python3 manage.py load_data

Data Generation

This sections describes how we generated data and how to generate data in same way.

  • The settings.yml file contains some necessary information for data generation. Please strictly follow the format and naming conventions.

    • For more information on the config file refer to the Buisness Rules section below.
    • The config["data_generation"]["aircrafts"] defines the number of aircrafts to be generated
      • Prn ,Flight counts probabilistic proportional to the number of aircrafts
      • For refference, we created 1100 aircrafts
      • Accordingly it generated 81k+ flight, 5.9M+ pnr and 13.6M+ passengers.
      • Causing 2.5GB+ of data ( takes 1-1:30 hours to generate )

  • Run the flight/management/data/generate.ipynb to generate some csv

    • This will generate the following files
      • flight/management/data/in-cabin.csv
      • flight/management/data/in-class.csv
      • flight/management/data/in-ssr.csv
      • flight/management/data/in-group.csv
      • flight/management/data/in-aircraft.csv

  • Table : airport

    • source : flight/management/commands/in-airport.csv
    • model : Airport
    • table_name: flight_airport
    • command:
    python3 manage.py populate --filename flight/management/data/in-airports.csv --delete

  • Table : Cabin Types, Class Types, SSRs, Groups

    • source :
      • flight/management/commands/in-cabin.csv,
      • flight/management/commands/in-class.csv,
      • flight/management/commands/in-ssr.csv,
      • flight/management/commands/in-group.csv
    • model : Cabin, Class, SSR, Group
    • table_name: flight_cabin, flight_class, flight_ssr, flight_group
    • command:
    python3 manage.py populate_baisc
    • For table cleanup use
    python3 manage.py populate_baic --clean

  • Table : Aircraft

    • model : Aircraft
    • command:
    python3 manage.py populate_aircraft --clean

  • Table : Flight, FlightScheuleDate, FlightScheule

    • model : Flight, FlightScheuleDate, FlightScheule
    • command:
    python3 manage.py populate_flight --clean

  • Table : PNR, PnrFlightMapping, PnrPassenger

    • model : PNR, FlightPnrMapping
    • table_name: flight_pnr, flight_flightpnrmapping
    • command:
    python3 manage.py populate_pnr
    • For table cleanup use
    python3 manage.py populate_pnr --clean

Note

for data ingestion use this order, first airports, then basic, routes, then pnr

Buisness Rules

Buiness rules are defined in settings.yml file ( or any other file specified in the --config argument of the allocate command )

Note: File extension must be .yml

Example config file

data_generation:
  safe: False # checks before inserting data in db if set True
  aircrafts: 1100
  carriers: 15
  flights_per_aircraft: 15
  flights_per_aircraft_margin: 3
  flight_schedule_prefix : "SCH-ZZ-"
  paid_service: 0.08
  paid_service_score: 200
  loyality_program_score: 2000
  loyality_program: 0.003 # PROBABILITY
  pax_score: 50
  ssr_score: 100
  default_ssr_probability: 0.001
search:
  max_hop : 2 # max limit 2 
  downgrade: True
  upgrade: True
  skip_quantum: False # reduce time ( tries to do the same thing in classical way ) when set True
  use_inventory: False # calculate on inventory instead of capacity
  use_cabin_only: True
  neighboring_search: True # searh using same state airport
  alpha : 4
  beta : 0.3
scores:
  connecting_score: 100
  default_ssr_score: 200
  default_scd1_score: 300
  default_scd2_score: 200
  cabins:
    F: 2000
    B: 1000
    P: 500
    E: 100
  loyalty:
    Silver: 1000
    Gold: 2000
    Platinum: 3000
  • There are 3 sections in the config file:
    • data_generation : contains the parameters for data generation
      • Only used in data generation, can be skipped for pnr allocation
      • All fields are compulsory
    • search : Contains the parameters for search or pnr reallocation
      • max_hop : maximum number of hops/connection allowed
      • downgrade : if set to True allows downgrading are allowed
      • upgrade : if set to True allows upgradings are allowed
      • skip_quantum : if set to True skips the quantum search and uses classical search
      • use_inventory : if set to True uses avilable inventory instead of avilable capacity
      • use_cabin_only : if set to True uses cabin only for search (If set to false pnrs should be mapped to class instead of cabin )
      • neighboring_search : if set to True uses neighboring airports for search, ( Neighboring airports are defined as airports in same state )
    • scores : This part is important for PNR scoring
      • connecting_score : score for connecting flights
      • default_ssr_score : score for default ssr
      • default_scd1_score : score for default scd1
      • default_scd2_score : score for default scd2
      • cabins : score for each cabin (edit the class.csv for class level scoring)
      • loyalty : score for each loyalty program

High Level Schema

High Level Schema

API Endpoints

Note

  • These api endpoints were used for our early development and testing.
  • Currently these apis are called internally (without http)
  • Still can be used for debugging and testing purposes.
  • Server must be running to use these apis.
  • Run python3 manage.py runserver to start the server.
  • Host address : http://127.0.0.1:8000/

API Doc

  • \canclled

    • Returns the list of cancelled flights
    • Method: GET
    • Filters by status = Cancelled
    • Sample Response:
    {
"count": 23,
"data": [
    {
        "flight_id": "INV-ZZ-4593004",
        "departure_airport": "AMD",
        "arrival_airport": "HYD",
        "status": "Cancelled",
        "arrival_time": "2024-05-21T13:37:00Z",
        "departure_time": "2024-05-21T04:59:00Z"
    }, ...
]
}

  • \pnr_ranking?flight_id={id}

    • Returns the list of passengers in the flight with the given flight_id
    • Method: GET
    • Filters by flight_id
    • Sample Response:
    {
"data": [
    {
        "pnr": "ZAYQ95",
        "score": 20000,
        "class": "R",
        "pax": 9
    },
    {
        "pnr": "BMHR27",
        "score": 16000,
        "class": "R",
        "pax": 7
    },...
]
}

  • \alt_flight?flight_id={id}

    • Returns the list of alternative flights for the given flight_id
    • Method: GET
    • Filters by flight_id and status = Scheduled
    • Sample Response:
    {
"data": [
    {
        "flight_id": "INV-ZZ-9565992",
        "departure_airport": "HYD",
        "arrival_airport": "AMD",
        "status": "Scheduled",
        "arrival_time": "2024-05-22T05:58:00Z",
        "departure_time": "2024-05-21T08:26:00Z",
        "total_avilable_seats": 160,
        "delay": "3:27:00",
        "flight_time": "21:32:00",
        "avilable_seats": {
            "C": 35,
            "R": 56,
            "E": 51,
            "F": 18
        },
        "n_score": 0
    },...
],
"r_flights": [
    {
        "flight_id": "INV-ZZ-4593004",
        "departure_airport": "AMD",
        "arrival_airport": "HYD",
        "status": "Cancelled",
        "arrival_time": "2024-05-21T13:37:00Z",
        "departure_time": "2024-05-21T04:59:00Z"
    }
],
"c_flights": [
    [
        {
            "flight_id": "INV-ZZ-6352197",
            "departure_airport": "AMD",
            "arrival_airport": "MAA",
            "status": "Scheduled",
            "arrival_time": "2024-05-22T16:15:00Z",
            "departure_time": "2024-05-22T13:07:00Z",
            "total_avilable_seats": 141,
            "delay": "1 day, 8:08:00",
            "flight_time": "3:08:00",
            "avilable_seats": {
                "C": 34,
                "R": 53,
                "E": 39,
                "F": 15
            }
        },
        {
            "flight_id": "INV-ZZ-4861936",
            "departure_airport": "MAA",
            "arrival_airport": "HYD",
            "status": "Scheduled",
            "arrival_time": "2024-05-23T18:28:00Z",
            "departure_time": "2024-05-23T08:12:00Z",
            "total_avilable_seats": 78,
            "delay": "15:57:00",
            "flight_time": "10:16:00",
            "avilable_seats": {
                "C": 19,
                "R": 29,
                "E": 22,
                "F": 8
            }
        }
    ],...
],
"t_flights": [
    [
        {
            "flight_id": "INV-ZZ-6352197",
            "departure_airport": "AMD",
            "arrival_airport": "MAA",
            "status": "Scheduled",
            "arrival_time": "2024-05-22T16:15:00Z",
            "departure_time": "2024-05-22T13:07:00Z",
            "total_avilable_seats": 141,
            "delay": "1 day, 8:08:00",
            "flight_time": "3:08:00",
            "avilable_seats": {
                "C": 34,
                "R": 53,
                "E": 39,
                "F": 15
            }
        },
        {
            "flight_id": "INV-ZZ-4277207",
            "departure_airport": "MAA",
            "arrival_airport": "BOM",
            "status": "Scheduled",
            "arrival_time": "2024-05-23T15:50:00Z",
            "departure_time": "2024-05-23T09:56:00Z",
            "total_avilable_seats": 34,
            "delay": "17:41:00",
            "flight_time": "5:54:00",
            "avilable_seats": {
                "C": 10,
                "R": 7,
                "E": 14,
                "F": 3
            }
        },
        {
            "flight_id": "INV-ZZ-6250134",
            "departure_airport": "BOM",
            "arrival_airport": "HYD",
            "status": "Scheduled",
            "arrival_time": "2024-05-24T05:44:00Z",
            "departure_time": "2024-05-23T16:54:00Z",
            "total_avilable_seats": 218,
            "delay": "1:04:00",
            "flight_time": "12:50:00",
            "avilable_seats": {
                "C": 50,
                "R": 72,
                "E": 68,
                "F": 28
            }
        }
    ],...
]
}
    • data : list of no-hop flights
    • r_flights : list of related cancelled flights ( solution space same )
    • c_flights : list of 1-hop flights
    • t_flights : list of 2-hop flights