def parse_header(header):
game_state = GameState()
base_featureset = FeatureSet()
for line in header.split("\n"):
line = line.strip()
if line.startswith("info,"):
try:
_, key, value = csv_split(line)
except Exception:
logging.error("Choked on line: %s" % line)
raise
if key in ["visteam", "hometeam"]:
setattr(GameState, key, value)
fs_key = "game_%s" % key
if fs_key in FeatureSet.__slots__:
setattr(FeatureSet, fs_key, value)
return game_state, base_featureset
def from_featureset_json(cls, json_str):
d = json.loads(json_str)
parse_map = FeatureSet.get_parse_map()
for k in d:
# all keys are strings in the .features format
if d[k] != UNK and k in parse_map:
try:
d[k] = parse_map[k](d[k])
except ValueError:
d[k] = UNK
obj = cls()
def unk_check(obj_key, keys=None):
""" If all specified keys are known, return True. Otherwise,
set UNK on the specified object key. """
if keys is None:
keys = [ obj_key ]
if all( (d[k] != UNK) for k in keys):
# proceed with calculation
return True
else:
setattr(obj, obj_key, UNK)
return False
obj.game_daynight = d["game_daynight"]
if unk_check("game_date"):
obj.game_month = cls.bucketized(d["game_date"].month, granularity=2)
if unk_check("game_date"):
obj.game_year = cls.bucketized(d["game_date"].year, granularity=5)
if unk_check("game_number"):
obj.game_number = cls.bucketized(d["game_number"], buckets=[0,1])
if unk_check("game_temp"):
obj.game_temp = cls.bucketized(d["game_temp"], granularity=5)
obj.game_site = d["game_site"]
obj.game_winddir = cls.get_winddir(d)
if unk_check("ab_inning"):
obj.ab_inning = cls.bucketized(d["ab_inning"], buckets=[1,4,7,10])
obj.ab_numballs = d["ab_numballs"]
obj.ab_numstrikes = d["ab_numstrikes"]
obj.ab_lrmatchup = cls.get_lrmatchup(d)
obj.batter_bats = d["batter_bats"]
obj.batter_fieldpos = d["batter_fieldpos"]
obj.batter_visorhome = d["batter_visorhome"]
if unk_check("batter_batpos"):
obj.batter_batpos = cls.bucketized(d["batter_batpos"], buckets=[1,3,5,8,10,11,12])
obj.label = d["label"]
for prefix in ["batter_", "pitcher_"]:
def key(k):
return prefix + k
def year_from_td(td):
return td.days / 365
if unk_check(key("age"), keys=[key("birthYear"), key("birthMonth"), key("birthDay"), "game_date"]):
birthday = datetime.datetime(d[key("birthYear")], d[key("birthMonth")], d[key("birthDay")])
setattr(obj, key("age"), cls.bucketized(year_from_td(d["game_date"] - birthday), granularity=3))
if unk_check(key("weight")):
setattr(obj, key("weight"), cls.bucketized(d[key("weight")], granularity=10))
if unk_check(key("height")):
setattr(obj, key("height"), cls.bucketized(d[key("height")], granularity=3))
setattr(obj, key("team"), d[key("team")])
setattr(obj, key("throws"), d[key("throws")])
setattr(obj, key("birthCountry"), d[key("birthCountry")])
if unk_check(key("experience"), keys=[key("debut"), "game_date"]):
setattr(obj, key("experience"), cls.bucketized(year_from_td(d["game_date"] - d[key("debut")]), granularity=3))
return obj
def copy(self):
# this is very puzzling; why does a new object created in the context of a given object have all the fields of the object in which the new object is created...?
new_obj = FeatureSet()
assert(new_obj.to_json() == self.to_json())
return new_obj