def mapper(stream):
fields = ['DayOfWeek', 'ArrDelay']
dow_map = [
'Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday',
'Sunday'
]
for dow, delay in mapred.iter_curated_fields(stream, fields):
dow_str = dow_map[int(dow) - 1]
mapred.send(dow_str, (delay, 1))
def mapper_step1(stream):
"""This mapper produces the keys to be used for the first problem reduction step: selecting
the single best flight for each day/day-part/origin/dest.
"""
fields = [
'FlightDate', 'Origin', 'Dest', 'ArrDelay', 'CRSDepTime',
'UniqueCarrier', 'FlightNum'
]
for date, origin, dest, delay, time, carrier, flight in mapred.iter_curated_fields(
stream, fields):
try:
minute_of_day = int(time[:2]) * 60 + int(time[2:])
except:
pass
else:
period = 'AM' if minute_of_day < 12 * 60 else 'PM'
mapred.send((date, period),
(date, origin, dest, delay, time, carrier, flight))
def mapper(stream):
fields = ['Origin', 'Dest', 'UniqueCarrier', 'ArrDelay']
for origin, dest, carrier, delay in mapred.iter_curated_fields(
stream, fields):
mapred.send((origin, dest, carrier), (delay, 1))
def mapper(stream):
fields = ['Origin', 'Dest', 'DepDelay']
for origin, dest, delay in mapred.iter_curated_fields(stream, fields):
mapred.send((origin, dest), (delay, 1))
def mapper(stream):
fields = ['Origin', 'Dest']
for origin, dest in mapred.iter_curated_fields(stream, fields):
mapred.send(origin, 1)
mapred.send(dest, 1)
def mapper(stream):
fields = ['UniqueCarrier', 'ArrDelay']
for carrier, delay in mapred.iter_curated_fields(stream, fields):
mapred.send(carrier, (delay, 1))