def __init__(self,
bins,
data_path,
calibration_features,
tt_lstm_config_path,
soccer_data_store_dir,
apply_old,
apply_difference,
focus_actions_list=[]):
self.bins = bins
# self.bins_names = bins.keys()
self.apply_old = apply_old
self.apply_difference = apply_difference
self.data_path = data_path
self.calibration_features = calibration_features
if self.apply_difference:
self.calibration_values_all_dict = {
'all': {
'cali_sum': [0],
'model_sum': [0],
'number': 0
}
}
else:
self.calibration_values_all_dict = {
'all': {
'cali_sum': [0, 0, 0],
'model_sum': [0, 0, 0],
'number': 0
}
}
self.soccer_data_store_dir = soccer_data_store_dir
self.tt_lstm_config = TTLSTMCongfig.load(tt_lstm_config_path)
self.focus_actions_list = focus_actions_list
if self.apply_difference:
self.save_calibration_dir = './calibration_results/difference-calibration-{0}-{1}.txt'. \
format(str(self.focus_actions_list), datetime.date.today().strftime("%Y%B%d"))
else:
self.save_calibration_dir = './calibration_results/calibration-{0}-{1}.txt'. \
format(str(self.focus_actions_list), datetime.date.today().strftime("%Y%B%d"))
self.save_calibration_file = open(self.save_calibration_dir, 'w')
if apply_difference:
self.teams = ['home-away']
else:
self.teams = ['home', 'away', 'end']
from td_three_prediction_two_tower_lstm_v_correct_dir.support.data_processing_tools import normalize_data
from td_three_prediction_two_tower_lstm_v_correct_dir.nn_structure.td_tt_lstm import td_prediction_tt_embed
from td_three_prediction_two_tower_lstm_v_correct_dir.support.plot_tools import compute_game_values, read_plot_model
if __name__ == '__main__':
data_store_dir = "/cs/oschulte/Galen/Hockey-data-entire/Hybrid-RNN-Hockey-Training-All-feature5-scale" \
"-neg_reward_v_correct__length-dynamic/"
data_path = "/cs/oschulte/Galen/Hockey-data-entire/Hockey-Match-All-data/"
tt_lstm_config_path = '../icehockey-config.yaml'
home_team = 'Penguins'
away_team = 'Canadiens'
target_game_id = str(1403)
dir_all = os.listdir(data_path)
game_name_dir = find_game_dir(dir_all, data_path, target_game_id)
tt_lstm_config = TTLSTMCongfig.load(tt_lstm_config_path)
learning_rate = tt_lstm_config.learn.learning_rate
if learning_rate == 1e-5:
learning_rate_write = 5
elif learning_rate == 1e-4:
learning_rate_write = 4
sess_nn = tf.InteractiveSession()
model_nn = td_prediction_tt_embed(
feature_number=tt_lstm_config.learn.feature_number,
home_h_size=tt_lstm_config.Arch.HomeTower.home_h_size,
away_h_size=tt_lstm_config.Arch.AwayTower.away_h_size,
max_trace_length=tt_lstm_config.learn.max_trace_length,
learning_rate=tt_lstm_config.learn.learning_rate,
embed_size=tt_lstm_config.learn.embed_size,