def mainKyle():
# Set parameters for loading the data
num_records = 'all'
csvfile = "Data/BKGDAT_ZeroTOTALBKD.txt"
# Set parameters for filtering the data
market = AirportCodes.London
orgs=[AirportCodes.Dubai, market]
dests=[AirportCodes.Dubai, market]
cabins=["Y"]
# Get the data, filter it, and group it by flight
print "Loading " + csvfile
f = FeatureFilter(num_records, csvfile)
print "Filtering"
data = f.getDrillDown(orgs=orgs, dests=dests, cabins=cabins)
print "Grouping by flight"
unique_flights = f.getUniqueFlights(data)
# Encode the flights
print "Encoding flight data"
start = -90
stop = 0
num_points = 31
interp_params = (start, stop, num_points)
bin_size = 3
date_reduction = 0
cat_encoding = (bin_size, date_reduction)
num_folds = 3
X, y, ids = encodeFlights(unique_flights, interp_params, cat_encoding)
_, num_features = X[0].shape
print 'Generating k-fold'
kf = kFoldSplit(X, y, ids, num_folds)
print 'Selecting features'
model = KNeighborsRegressor
print sequentialForwardFeatureSelection(model, kf, num_features)
def mainRyan():
# Set parameters for loading the data
num_records = 'all'
csvfile = "Data/BKGDAT_ZeroTOTALBKD.txt"
# Set parameters for filtering the data
market = AirportCodes.Frankfurt
cabins=["Y"]
# Get the data, filter it, and group it by flight
print "Loading " + csvfile
f = FeatureFilter(num_records, csvfile)
print "Filtering"
if market is None:
orgs=[AirportCodes.Dubai, AirportCodes.London, AirportCodes.Bahrain, AirportCodes.Frankfurt, AirportCodes.Bangkok]
dests=[AirportCodes.Dubai, AirportCodes.London, AirportCodes.Bahrain, AirportCodes.Frankfurt, AirportCodes.Bangkok]
data = f.getDrillDown(orgs=orgs, dests=dests, cabins=cabins)
else:
orgs=[AirportCodes.Dubai, market]
dests=[AirportCodes.Dubai, market]
data = f.getDrillDown(orgs=orgs, dests=dests, cabins=cabins)
print "Grouping by flight"
unique_flights = f.getUniqueFlights(data)
# Encode the flights
print "Encoding flight data"
start = -90
stop = 0
num_points = 31
interp_params = (start, stop, num_points)
bin_size = 3
date_reduction = 0
cat_encoding = (bin_size, date_reduction)
X, y, ids = encodeFlights(unique_flights, interp_params, cat_encoding)
X_train, y_train, X_test, y_test, ids_train, ids_test = aggregateTrainTestSplit(X, y, ids, 0.75)
return X_train, y_train, X_test, y_test, ids_train, ids_test, interp_params, cat_encoding