parser.add_argument('--cn')
parser.add_argument('--mn')
parser.add_argument('--mt')
parser.add_argument('--cf')
args = parser.parse_args()
collection_name = args.cn
model_name = args.mn
model_type = args.mt
config_path = args.cf
config = ConfigParser.ConfigParser()
config.read(config_path)
root_logger.info('Get train data.')
data_rep = DataRepository(uri=config.get('defaults', 'MONGO_URI'))
X = []
Y = []
for row in data_rep.get(collection_name, { '$or': [{'Y': [1]}, {'Y': [0]}]}):
root_logger.info(row['meta'])
X.append(row['X'])
Y.append(row['Y'])
root_logger.info('We have {0} rows to train'.format(len(Y)))
root_logger.info('Train model.')
model = StatModelFactory.create(model_type)
model.train(X, Y)
model_storage = AzureBlobStorage(
config_path = args.cf
source_type = args.st
config = ConfigParser.ConfigParser()
config.read(config_path)
# (stat_season, games_season)
seasons = [
(2012, 2012),
(2013, 2013),
(2014, 2014),
(2015, 2015),
(2015, 2016),
]
data_rep = DataRepository(uri=config.get('defaults', 'MONGO_URI'))
for season in seasons:
root_logger.info('stat_season: {0}, games_season: {1}'
.format(season[0], season[1]))
ds = SportReferenceDataSource(
base_url=__source_type_base_url_map[source_type],
team_stat_season=season[0],
games_season=season[1],
game_type=source_type,
row_parse_strategy=__source_type_map[source_type],
cache_team_stats=True)
X, Y, metadata = ds.load(dict(date_from=date_from, date_to=date_to))
date_from = date_utils.try_parse(args.df)
date_to = date_utils.try_parse(args.dt)
source_name = args.sn
source_type = args.st
config = ConfigParser.ConfigParser()
config.read('scripts.cfg')
ds = SportReferenceDataSource(
base_url=__source_type_base_url_map[source_type],
team_stat_season=2015,
games_season=2016,
game_type=source_type,
row_parse_strategy=__source_type_map[source_type],
cache_team_stats=True)
X, Y, metadata = ds.load(dict(date_from=date_from, date_to=date_to))
data_rep = DataRepository(uri=config.get('defaults', 'MONGO_URI'))
meta_keys = metadata[0].keys()
for i in range(len(Y)):
if not Y[i]:
continue
meta_data_row = [metadata[i][k] for k in metadata[i].keys()]
res = data_rep.update(
source_name,
{ 'meta.{0}'.format(meta_keys[j]):meta_data_row[j] for j in range(len(meta_keys)) },
{ 'Y': [Y[i]] }
)
root_logger.info(res)
