def date_diff(warn_date, gsr_date):
"""
Computes date difference based on warning date and event date.
:param warn_date: Event_Date in the warning
:param gsr_date: Event_Date in the GSR
:return: integer, negative values mean the warning was later than the event
"""
gsr_d = parse(gsr_date).date()
warn_d = parse(warn_date).date()
date_diff = (gsr_d - warn_d).days
return date_diff
def score_one(warn_,
event_,
max_dist=Defaults.MAX_DIST,
dist_buffer=Defaults.DIST_BUFFER,
max_date_diff=Defaults.MAX_DATE_DIFF,
legit_actors=ACTORS,
wildcards=WILDCARD_ACTORS,
ls_weight=Defaults.LS_WEIGHT,
ds_weight=Defaults.DS_WEIGHT,
as_weight=Defaults.AS_WEIGHT,
ess_weight=Defaults.ESS_WEIGHT):
"""
Scores a single warning against a single event
:param warn_: Dict with data for a warning
:param event_: Dict with data for an event
:return: Dict with scoring details
"""
bad_qs = False
error_list = []
notice_list = []
if ls_weight < 0:
bad_qs = True
error_list.append("LS Weight must be positive")
if ds_weight < 0:
bad_qs = True
error_list.append("DS Weight must be positive")
if as_weight < 0:
bad_qs = True
error_list.append("AS Weight must be positive")
if ess_weight < 0:
bad_qs = True
error_list.append("ESS Weight must be positive")
weight_sum = ls_weight + ds_weight + as_weight + ess_weight
if weight_sum != 4.0:
notice_list.append("Reweighting so that sum of weights is 4.0")
ls_weight = 4 * ls_weight / weight_sum
ds_weight = 4 * ds_weight / weight_sum
as_weight = 4 * as_weight / weight_sum
ess_weight = 4 * ess_weight / weight_sum
out_dict = dict()
# Compute the distance
out_dict[JSONField.WARNING_ID] = warn_[JSONField.WARNING_ID]
out_dict[JSONField.EVENT_ID] = event_[JSONField.EVENT_ID]
if error_list:
out_dict["Errors"] = error_list
else:
warn_lat = warn_[JSONField.LATITUDE]
warn_long = warn_[JSONField.LONGITUDE]
event_lat = event_[JSONField.LATITUDE]
event_long = event_[JSONField.LONGITUDE]
loc_approx = event_[JSONField.APPROXIMATE_LOCATION]
out_dict[JSONField.APPROXIMATE_LOCATION] = loc_approx
dist = distance((warn_lat, warn_long), (event_lat, event_long)).km
out_dict["Distance"] = dist
warn_event_date = warn_[JSONField.EVENT_DATE]
event_event_date = event_[JSONField.EVENT_DATE]
gsr_d = parse(event_event_date).date()
warn_d = parse(warn_event_date).date()
date_diff = (gsr_d - warn_d).days
date_diff = np.abs(date_diff)
out_dict["Date Difference"] = date_diff
ls = MaScorer.location_score(dist, loc_approx, max_dist,
dist_buffer)
out_dict[ScoreComponents.LS] = ls
date_delta = Scorer.date_diff(warn_event_date, event_event_date)
ds = Scorer.date_score(date_delta, max_date_diff)
out_dict[ScoreComponents.DS] = ds
# Event Subtype
warn_es = warn_[JSONField.SUBTYPE]
gsr_es = event_[JSONField.SUBTYPE]
ess = MaScorer.event_subtype_score(warn_es, gsr_es)
out_dict[ScoreComponents.ESS] = ess
# Actor Score
warn_actor = warn_[JSONField.ACTOR]
event_actor = event_[JSONField.ACTOR]
_as = MaScorer.actor_score(warn_actor, event_actor, legit_actors,
wildcards)
out_dict[ScoreComponents.AS] = _as
if min(ls, ds) == 0:
qs = 0
else:
qs = ls_weight * ls + ds_weight * ds + as_weight * _as + ess_weight * ess
out_dict[ScoreComponents.QS] = qs
# Quality Score
if notice_list:
out_dict["Notices"] = notice_list
return out_dict