def query():
if file_access(GAME_RESULTS_PATH) and file_access(TEST_GAMES_PATH):
with open(GAME_RESULTS_PATH) as file:
games = json.load(file)
with open(TEST_GAMES_PATH) as file:
test_games = json.load(file)
return games, test_games
games = []
for year in years:
for week in weeks:
headers = {
'accept': 'application/json',
}
params = (
('year', str(year)),
('week', str(week)),
('seasonType', 'regular'),
)
response = requests.get('https://api.collegefootballdata.com/games', headers=headers, params=params)
game = response.json()
games.append(game)
test_games = []
if test_year != 0:
for week in weeks:
headers = {
'accept': 'application/json',
}
params = (
('year', str(test_year)),
('week', str(week)),
('seasonType', 'regular'),
)
response = requests.get('https://api.collegefootballdata.com/games', headers=headers, params=params)
game = response.json()
test_games.append(game)
with open(TEST_GAMES_PATH, 'w') as f:
json.dump(test_games, f)
with open(GAME_RESULTS_PATH, 'w') as f:
json.dump(games, f)
return games, test_games
def stat_files_exist():
for statistic in statistics:
for year in years:
stat_file = STAT_PATH + statistic + str(year) + ".json"
if not file_access(stat_file):
return False
return True
def main():
if file_access(GAME_DATA_PATH) and file_access(TEST_GAME_DATA_PATH):
print("Loading existing game data")
game_data = load(GAME_DATA_PATH)
test_game_data = load(TEST_GAME_DATA_PATH)
else:
print("Running queries to download game data...")
os.makedirs("data", exist_ok=True)
print("Loading all games in the specified range...")
games, test_games = games_query.query()
# talent = talent_query.query()
print("Finding dates of games...")
dates_to_games = retreive_all_dates(games)
print("Loading historical polls...")
ap_polls, coaches_polls, pred_polls = polls_query.query(dates_to_games)
print("Loading historical statistics...")
stats_query.query(dates_to_games)
print("Combining and normalizing game data...")
game_data = combine_game_data(dates_to_games, ap_polls, coaches_polls, pred_polls)
game_data = normalize(game_data)
print("Saving data to", GAME_DATA_PATH)
save(game_data, GAME_DATA_PATH)
# Repeat all that for test seasons
print("Doing that all again for test seasons...")
dates_to_games = retreive_all_dates(test_games)
ap_polls, coaches_polls, pred_polls = polls_query.query(dates_to_games, append=True)
stats_query.query(dates_to_games, append=True)
test_game_data = combine_game_data(dates_to_games, ap_polls, coaches_polls, pred_polls)
test_game_data = normalize(test_game_data)
save(test_game_data, TEST_GAME_DATA_PATH)
# game_data = game_data[:1000]
all_inputs, all_outputs = create_netdata_from_gamedata(game_data)
test_inputs, test_outputs = create_netdata_from_gamedata(test_game_data)
# train_neat(all_inputs, all_outputs)
print("Doing NN stuff...")
# scikit_net(all_inputs, all_outputs, test_inputs, test_outputs)
tf_net(all_inputs, all_outputs, test_inputs, test_outputs)
def query(dates_to_games, append=False):
for statistic in statistics:
game_years = set(
[year_from_date(date) for date in dates_to_games.keys()])
for year in game_years:
stat_file = STAT_PATH + statistic + str(year) + ".json"
if append == True or not file_access(stat_file):
stats = {}
games_in_year = [(date, dates_to_games[date])
for date in dates_to_games.keys()
if year_from_date(date) == str(year)]
for date, games in games_in_year:
teams = teams_in_games(games)
stats = update_stats(stats, statistic, date, teams)
stat_file = STAT_PATH + statistic + str(year) + ".json"
os.makedirs(os.path.dirname(stat_file), exist_ok=True)
with open(stat_file, 'w+') as f:
json.dump(stats, f)
def train_net(inputs, outputs, test_inputs, test_outputs, load=False):
inputs = np.asarray(inputs)
outputs = np.asarray(outputs)
test_inputs = np.asarray(test_inputs)
if load and file_access(KERAS_MODEL_PATH):
print("Loading model from", KERAS_MODEL_PATH)
model = load_model(KERAS_MODEL_PATH)
else:
model = create_uniform_model(inputs.shape[1])
# early_stopping = keras.callbacks.EarlyStopping(monitor='val_loss',
# min_delta=5 * 10 ** -6, patience=20,
# verbose=0, mode='auto',
# baseline=None,
# restore_best_weights=True)
# model.fit(x=inputs, y=outputs,
# validation_data=(val_in, val_out),
# epochs=1000, callbacks=[early_stopping], verbose=1)
print(model.summary())
model.fit(x=inputs, y=outputs, epochs=1000, verbose=1)
return model, model.predict(x=test_inputs)
def query(dates_to_games, append=False):
'''
Query online historical poll data to build a database of team ranking history
:param dates_to_games: dictionary using a date as keys to all games played on that date
:param append: Add to exiting data?
:return: Three dictionaries: ap_polls, coaches_polls, pred_polls
'''
# Check if we already have all the poll data
if append == False and file_access(PATH_AP_JSON) and file_access(
PATH_COACHES_JSON) and file_access(PATH_ALL_JSON):
with open(PATH_AP_JSON) as file:
ap_polls = json.load(file)
with open(PATH_COACHES_JSON) as file:
coaches_polls = json.load(file)
with open(PATH_ALL_JSON) as file:
pred_polls = json.load(file)
return ap_polls, coaches_polls, pred_polls
ap_polls = {}
coaches_polls = {}
pred_polls = {}
game_years = set([year_from_date(date) for date in dates_to_games.keys()])
for year in game_years:
for week in weeks:
# Query for ap and coaches poll data from api.collegefootballdata.com
headers = {
'accept': 'application/json',
}
params = (('year', str(year)), ('week', str(week)))
response = requests.get(
'https://api.collegefootballdata.com/rankings',
headers=headers,
params=params)
ap_poll_query = response.json()[0]['polls'][1]['ranks']
coaches_poll_query = response.json()[0]['polls'][0]['ranks']
ap_poll = {}
coaches_poll = {}
for pos in ap_poll_query:
team_name = pos['school']
rank = pos['rank']
ap_poll[team_name] = rank
for pos in coaches_poll_query:
team_name = pos['school']
rank = pos['rank']
coaches_poll[team_name] = rank
# Store the polls for every date a game was played
ap_polls[str(year) + "," + str(week)] = ap_poll
coaches_polls[str(year) + "," + str(week)] = coaches_poll
for date, games in dates_to_games.items():
teams = teams_in_games(games)
pred_polls = update_pred_polls(pred_polls, date, teams)
if append:
write = "w+"
else:
write = "w"
with open(PATH_AP_JSON, write) as f:
json.dump(ap_polls, f)
with open(PATH_COACHES_JSON, write) as f:
json.dump(coaches_polls, f)
with open(PATH_ALL_JSON, write) as f:
json.dump(pred_polls, f)
return ap_polls, coaches_polls, pred_polls