def oneHotBookingClass(bc, bin_size=1):
""" Returns a binned 1-to-K or one-hot encoding of BC.
bc: a booking class letter
bin_size: number of bc that fit into one bin
if bin_size=1 (default), we have a true 1-K encoding
"""
assert len(Utils.bc_hierarchy) % bin_size == 0, "Invalid Bin Size"
cabin, rank = Utils.mapBookingClassToCabinHierarchy(bc)
enc_vector = np.zeros(len(Utils.bc_hierarchy)/bin_size)
enc_vector[rank/bin_size] = 1
return enc_vector
def countCabinCapacityPerFlight(self):
"""
Counts the total capcity of a flight in every cabin on the plane
returns:
dictionary of {flight, dictionary of {cabin, cabin capacity}}
"""
flights = self.f.getUniqueFlightsAndBookings()
capacities = {}
for booking_group, data in flights:
flight = booking_group[0:4]
bc = booking_group[4]
cabin, rank = Utils.mapBookingClassToCabinHierarchy(bc)
if flight not in capacities:
capacities[flight] = {}
capacities[flight][cabin] = data['CAP'].mean()
return capacities
def countTotalBookedPerFlight(self):
"""
Counts the total number of passengers on a flight in every cabin on the
plane
returns:
dictionary of {flight, dictionary of {cabin, total booked}}
"""
flights = self.f.getUniqueFlightsAndBookings()
total_bookings = {}
for booking_group, data in flights:
flight = booking_group[0:4]
bc = booking_group[4]
cabin, rank = Utils.mapBookingClassToCabinHierarchy(bc)
if flight not in total_bookings:
total_bookings[flight] = {}
total_bookings[flight][cabin] = data['TOTALBKD'].mean()
return total_bookings
including OneHot or 1-K Methods and binning the rank
"""
return oneHotBookingClass(bc, bin_size)
def oneHotBookingClass(bc, bin_size=1):
""" Returns a binned 1-to-K or one-hot encoding of BC.
bc: a booking class letter
bin_size: number of bc that fit into one bin
if bin_size=1 (default), we have a true 1-K encoding
"""
assert len(Utils.bc_hierarchy) % bin_size == 0,
"Error: Bin Size must evenly divide into the number of booking classes"
cabin, rank = Utils.mapBookingClassToCabinHierarchy(bc)
enc_vector = np.zeros(len(Utils.bc_hierarchy)/bin_size)
enc_vector[rank/bin_size] = 1
return enc_vector
def sortBCGroupby(groupby):
tups = [(bc, bc_df) for bc, bc_df in groupby]
return sorted(tups, key=lambda tup: Utils.compareBCs(tup[0]))
def interpolate(keyday_vals, keydays, *args):
interps = [np.interp(keyday_vals, keydays, arg, left=0) for arg in args]
return interps
