test.py

import learning, util, sys, DataExtractor
from learning import *
from DataExtractor import DataExtractor
learner = StochasticGradientLearner(footballFeatureExtractor)

data = list()

for i in xrange(5, 13):
  data.append(DataExtractor(i).featureDictionary)
  
train = dict()

test = dict()

for i in xrange(0, 4):
  train.update(data[i])

for i in xrange(4, 8):
  test.update(data[i])


from optparse import OptionParser
parser = OptionParser()
def default(str):
  return str + ' [Default: %default]'
parser.add_option('-f', '--featureExtractor', dest='featureExtractor', type='string',
                  help=default('Which feature extractor to use (basic or custom)'), default="basic")
parser.add_option('-l', '--loss', dest='loss', type='string',
                  help=default('Which loss function to use (logistic, hinge, or squared)'), default="logistic")
parser.add_option('-i', '--initStepSize', dest='initStepSize', type='float',
                    help=default('the initial step size'), default=0.00001)
parser.add_option('-s', '--stepSizeReduction', dest='stepSizeReduction', type='float',
                    help=default('How much to reduce the step size [0, 1]'), default=1)
parser.add_option('-R', '--numRounds', dest='numRounds', type='int',
                    help=default('Number of passes over the training data'), default=100)
parser.add_option('-r', '--regularization', dest='regularization', type='float',
                    help=default('The lambda in L2 regularization'), default=0)
#parser.add_option('-d', '--dataset', dest='dataset', type='string',
#help=default('Prefix of dataset to load (files are <prefix>.{train,validation}.csv)'), default='toy')
parser.add_option('-v', '--verbose', dest='verbose', type='int',
                    help=default('Verbosity level'), default=0)

options, extra_args = parser.parse_args(sys.argv[1:])
if len(extra_args) != 0:
  print "Ignoring extra arguments:", extra_args

if options.loss == 'logistic':
  loss = logisticLoss
  lossGradient = logisticLossGradient
elif options.loss == 'hinge':
  loss = hingeLoss
  lossGradient = hingeLossGradient
elif options.loss == 'squared':
  loss = squaredLoss
  lossGradient = squaredLossGradient
else:
  print "Invalid loss function"

learner.learn(train.values(), test.values(), loss, lossGradient, options)