def compute_distribution(self, predict_season: int, latest_episode: int,
train_seasons: Set[int]) -> Dict[Player, float]:
equal_layer = EqualLayer()
new_distribution = equal_layer.compute_distribution(
predict_season, latest_episode, train_seasons)
for layer in self.__layers:
layer_distribution = layer.compute_distribution(
predict_season, latest_episode, train_seasons)
for player, likelihood in new_distribution.items():
new_distribution[
player] = layer_distribution[player] * likelihood
return new_distribution
def compute_distribution(
self, predict_season: int, latest_episode: int,
train_seasons: Set[int]) -> Dict[Player, np.array]:
all_predictions = self.__layer.predict(predict_season, latest_episode,
train_seasons)
if not all_predictions:
return EqualLayer().compute_distribution(predict_season,
latest_episode,
train_seasons)
distribution = dict()
for player, predict in all_predictions.items():
if predict.exclusion:
distribution[player] = 0.0
else:
total_likelihood = 1.0
for likelihood in predict.predictions:
total_likelihood *= likelihood
distribution[player] = total_likelihood
if self.__smoothness:
for player, likelihood in distribution.items():
number_predictions = len(all_predictions[player].predictions)
distribution[player] = likelihood**self.get_smoothness_factor(
number_predictions)
return distribution
def compute_distribution(self, predict_season: int, latest_episode: int, train_seasons: Set[int]) -> Dict[Player, float]:
available_seasons = self.seasons_with_data()
if predict_season not in available_seasons:
return EqualLayer().compute_distribution(predict_season, latest_episode, train_seasons)
train_seasons = train_seasons.intersection(available_seasons)
extractor = WikipediaExtractor(predict_season, train_seasons, self.__pca_components, self.__unlikely_z_score,
self.__random_generator)
extractor.train()
return self.__predict(extractor)
def compute_distribution(self, predict_season: int, latest_episode: int,
train_seasons: Set[int]) -> Dict[Player, float]:
if predict_season in SUSPICION_DATA:
exclude_players = SUSPICION_DATA[predict_season]
distribution = dict()
for player in get_players_in_season(predict_season):
distribution[
player] = self.__EPSILON if player in exclude_players else 1.0
return distribution
else:
return EqualLayer().compute_distribution(predict_season,
latest_episode,
train_seasons)
def compute_distribution(self, predict_season: int, latest_episode: int, train_seasons: Set[int]) -> Dict[Player, float]:
available_seasons = EXAM_DATA.keys()
train_seasons = train_seasons.intersection(available_seasons)
if predict_season not in available_seasons:
return EqualLayer().compute_distribution(predict_season, latest_episode, train_seasons)
estimator = self.__train(train_seasons)
alive_players = EXAM_DATA[predict_season].get_alive_players(latest_episode)
result = {player: 1.0 if player in alive_players else 0.0 for player in get_players_in_season(predict_season)}
for episode in EXAM_DATA[predict_season].episodes.values():
if episode.id > latest_episode or episode.result.drop != DropType.EXECUTION_DROP:
continue
prediction = self.__predict_for_episode(episode, alive_players, estimator)
for player, likelihood in prediction.items():
result[player] *= likelihood
return result