File size: 1,915 Bytes
d6ea71e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 |
import pandas as pd
import pytest
from socceraction.vaep import VAEP
from socceraction.vaep import features as fs
@pytest.fixture(scope="session")
def vaep_model(sb_worldcup_data: pd.HDFStore) -> VAEP:
# Test the vAEP framework on the StatsBomb World Cup data
model = VAEP(nb_prev_actions=1)
# comppute features and labels
games = sb_worldcup_data["games"]
features = pd.concat(
[
model.compute_features(game, sb_worldcup_data[f"actions/game_{game.game_id}"])
for game in games.iloc[:-1].itertuples()
]
)
expected_features = set(fs.feature_column_names(model.xfns, model.nb_prev_actions))
assert set(features.columns) == expected_features
labels = pd.concat(
[
model.compute_labels(game, sb_worldcup_data[f"actions/game_{game.game_id}"])
for game in games.iloc[:-1].itertuples()
]
)
expected_labels = {"scores", "concedes"}
assert set(labels.columns) == expected_labels
assert len(features) == len(labels)
# fit the model
model.fit(features, labels)
return model
@pytest.mark.e2e
def test_predict(sb_worldcup_data: pd.HDFStore, vaep_model: VAEP) -> None:
games = sb_worldcup_data["games"]
game = games.iloc[-1]
actions = sb_worldcup_data[f"actions/game_{game.game_id}"]
ratings = vaep_model.rate(game, actions)
expected_rating_columns = {"offensive_value", "defensive_value", "vaep_value"}
assert set(ratings.columns) == expected_rating_columns
@pytest.mark.e2e
def test_predict_with_missing_features(sb_worldcup_data: pd.HDFStore, vaep_model: VAEP) -> None:
games = sb_worldcup_data["games"]
game = games.iloc[-1]
actions = sb_worldcup_data[f"actions/game_{game.game_id}"]
X = vaep_model.compute_features(game, actions)
del X["period_id_a0"]
with pytest.raises(ValueError):
vaep_model.rate(game, actions, X)
|