Discounting a player

[mildfuzz]

Fairly certain Harry Kane is going to leave, but my prediction keeps picking him. Is there a way to discount a player from the model?

[jack89roberts]

That's not functionality we have at the moment, but you can try this as a workaround and it's something we can consider adding in the future:

Save this as a file delete_points.py on your system:

import argparse

from airsenal.framework.schema import PlayerPrediction, session
from airsenal.framework.utils import get_latest_prediction_tag, get_player


def delete_points(player):
    print(f"Deleting latest predictions for {player}")
    tag = get_latest_prediction_tag()
    preds = (
        session.query(PlayerPrediction)
        .filter_by(tag=tag, player_id=player.player_id)
        .all()
    )
    for p in preds:
        p.predicted_points = 0

    session.commit()
    print("Done!")


def main():
    parser = argparse.ArgumentParser(
        description="Delete points predictions for a player"
    )
    parser.add_argument("player", help="Player name")
    args = parser.parse_args()
    player = get_player(args.player)
    delete_points(player)


if __name__ == "__main__":
    main()

Then follow these steps (from the directory you saved the delete_points.py file):

conda activate airsenalenv
airsenal_update_db
airsenal_run_prediction
python delete_points.py "Harry Kane"
airsenal_make_squad

You can vary the arguments to the prediction and make squad steps as normal. The delete_points.py script just sets all the predicted points for a player to zero in the database (could also run it multiple times with a different player each time if you wanted).