def test_estimate_class_level_struct(self):
utilities = {'fare1': -2.8564, 'fare2': -2.5684}
probs, nofly_prob = mfrm.selection_probs(utilities, 0.5)
estimations = mfrm.estimate_class_level({
'fare1': 3,
'fare2': 0
}, {
'fare1': 1,
'fare2': 0.
}, probs, nofly_prob)
self.assertIsInstance(estimations, dict)
self.assertEqual(len(estimations), 2)
self.assertEqual(sorted(estimations.keys()), ['fare1', 'fare2'])
self.assertEqual(sorted(estimations['fare1'].keys()),
['demand', 'recapture', 'spill'])
def test_estimate_class_level_ex1(self):
utilities = {'fare1': -2.8564, 'fare2': -2.5684}
probs, nofly_prob = mfrm.selection_probs(utilities, 0.5)
estimations = mfrm.estimate_class_level({
'fare1': 3,
'fare2': 0
}, {
'fare1': 1,
'fare2': 0.
}, probs, nofly_prob)
self.assertAlmostEqual(estimations['fare1']['spill'], 0, places=2)
self.assertAlmostEqual(estimations['fare1']['recapture'], 0.86, 2)
self.assertAlmostEqual(estimations['fare1']['demand'], 2.14, 2)
self.assertAlmostEqual(estimations['fare2']['spill'], 2.86, places=2)
self.assertAlmostEqual(estimations['fare2']['recapture'], 0, 2)
self.assertAlmostEqual(estimations['fare2']['demand'], 2.86, 2)
def test_estimate_paper_ex1(self):
"""This test is based on the exaple #1 in the paper"""
utilities = {'fare1': -2.8564, 'fare2': -2.5684}
probs, nofly_prob = mfrm.selection_probs(utilities, 0.5)
estimations = mfrm.estimate_host_level({
'fare1': 3,
'fare2': 0
}, {
'fare1': 1,
'fare2': 0.
}, probs, nofly_prob)
self.assertIsInstance(estimations, tuple)
self.assertEqual(len(estimations), 3)
# round numbers
estimations = round_tuple(estimations)
self.assertTupleEqual(estimations, (5, 2.86, 0.86))
def test_estimate_class_level_regression_2(self):
utilities = {
'25169088_1990': -0.32789022,
'25177238_2490': 0.74647802,
'30920878_3590': 1.1093128,
'30921208_3590': -0.95266426,
'25176068_2990': 0.60245919,
'27334408_3590': 1.3004869,
'27336048_2490': -0.72204816,
'25176068_2490': 1.1348069,
'27330978_3950': -0.95266426,
'27331818_2490': 1.1348069,
'25174968_2990': -0.95266426,
'27331818_3590': 0.60245919,
'25172878_2990': 1.3004869,
'27330978_3590': -0.95266426,
'25174968_3590': -0.95266426,
'27334408_3950': 1.3004869,
'30920428_2490': -0.72204816,
'27331818_2990': 0.60245919,
'25168168_1990': -0.86685419,
'25168168_3590': 0.028352857,
'25177238_2990': 0.37813818,
'25170958_1990': 0.43109083,
'25178118_1500': 1.557166,
'25172878_3590': 1.3004869,
'25173528_2990': -0.86880505,
'25173528_2490': 0.77468485,
'27332988_2990': 0.59399945,
'25168168_2490': -0.72204816,
'25170958_2490': 0.5758968,
'27332988_2490': -0.18308425,
'25168168_2990': 0.028352857
}
observed = {
'25169088_1990': 2,
'25177238_2490': 5,
'25176068_2990': 1,
'25177238_2990': 1,
'25170958_1990': 7,
'25176068_2490': 5,
'25168168_1990': 1,
'25174968_2990': 2,
'25173528_2990': 2,
'25173528_2490': 1,
'25172878_2990': 3,
'25168168_2490': 3,
'25170958_2490': 3,
'25178118_1500': 5
}
availability = {
'25177238_2490': 0.954545454545455,
'25169088_1990': 1.0,
'30920878_3590': 1.0,
'30921208_3590': 1.0,
'25176068_2990': 0.272727272727273,
'27334408_3590': 0.409090909090909,
'27334408_3950': 0.590909090909091,
'25176068_2490': 0.909090909090909,
'27331818_2990': 0.363636363636364,
'27331818_2490': 0.590909090909091,
'25172878_3590': 0.136363636363636,
'27331818_3590': 0.136363636363636,
'25172878_2990': 0.909090909090909,
'27330978_3590': 0.454545454545455,
'25174968_3590': 0.0454545454545455,
'27336048_2490': 1.0,
'30920428_2490': 1.0,
'27330978_3950': 0.590909090909091,
'25168168_1990': 0.727272727272727,
'25168168_3590': 0.0454545454545455,
'25177238_2990': 0.0909090909090909,
'25170958_1990': 0.590909090909091,
'25178118_1500': 1.0,
'25174968_2990': 1.0,
'25173528_2990': 0.590909090909091,
'25173528_2490': 0.5,
'27332988_2990': 0.545454545454545,
'25168168_2490': 0.454545454545455,
'25170958_2490': 0.454545454545455,
'27332988_2490': 0.681818181818182,
'25168168_2990': 0.0454545454545455
}
probs, nofly_prob = mfrm.selection_probs(utilities, 0.7)
class_level = mfrm.estimate_class_level(observed, availability, probs,
nofly_prob)
self.assertEqual(class_level['25168168_2990'], {
'demand': 0,
'spill': 0,
'recapture': 0
})
total_demand = sum([round(d['demand']) for d in class_level.values()])
self.assertEqual(total_demand, 50)
total_demand = sum([round(d['spill']) for d in class_level.values()])
self.assertEqual(total_demand, 20)
def test_estimate_class_level_regression_1(self):
utilities = {
'29255588_3950': 1.4659572,
'27330948_3950': 2.16431,
'29255588_2490': 1.1630461,
'29255578_2990': 1.3300509,
'29255508_3950': 0.43902999,
'29255578_3590': 0.83872116,
'29255578_3950': 0.70265454,
'29255528_3590': 0.52609205,
'29255518_3590': 0.52609205,
'30920848_3950': -0.19642138,
'27331928_3950': 0.096954226,
'27337358_3590': 0.52609205,
'27334478_3590': -0.12226862,
'29255548_3950': 1.4128097,
'29255558_3590': 0.2330209,
'29255588_2990': 0.99219722,
'29255538_3590': 0.76555932,
'27341178_3590': 0.61577702,
'29255548_3590': 1.3927615,
'29255558_2990': 0.72435057
}
observed = {
'29255588_3950': 3,
'27330948_3950': 1,
'29255578_2990': 7,
'27331928_2450': 1,
'27331928_3200': 4,
# missing in 'utilities'
'27331928_2490': 1,
'29255588_3018': 1,
'29255588_2490': 6,
'29255518_3950': 1,
'29255578_3018': 1,
'27331928_3950': 2,
'29255578_3950': 3,
'29255548_3950': 2,
'29255588_2518': 1,
'29255538_3590': 1,
'29255578_3590': 1,
'29255588_2241': 1
}
availability = {
'29255588_3950': 0.650909090909091,
'27330948_3950': 0.634146341463415,
'29255578_2990': 0.1875,
'27331928_2450': 0.00436681222707424,
'30920848_3950': 0.5,
'29255538_3950': 0.0411764705882353,
'29255528_3590': 0.796875,
'29255518_3950': 0.032967032967033,
'29255588_3590': 0.0254545454545455,
'29255518_3590': 0.648351648351648,
'29255558_3950': 0.00436681222707424,
'27334478_3590': 0.753246753246753,
'27334478_3950': 0.0779220779220779,
'29255528_3950': 0.0703125,
'29255548_3950': 0.521951219512195,
'29255558_3590': 0.148471615720524,
'29255578_3018': 0.00436681222707424,
'27331928_2490': 0.00436681222707424,
'29255548_3590': 0.268292682926829,
'29255588_2241': 0.00436681222707424,
'29255508_3950': 0.714285714285714,
'27331928_3200': 0.00436681222707424,
'29255588_2518': 0.00436681222707424,
'29255588_3018': 0.00436681222707424,
'29255588_2490': 0.0872727272727273,
'27331928_3950': 0.370517928286853,
'27337358_3590': 0.774193548387097,
'29255578_3950': 0.213235294117647,
'29255578_3590': 0.132352941176471,
'29255588_2990': 0.236363636363636,
'29255538_3590': 0.911764705882353,
'27341178_3590': 0.971428571428571,
'29255558_2990': 0.152838427947598
}
probs, nofly_prob = mfrm.selection_probs(utilities, 0.7)
class_level = mfrm.estimate_class_level(observed, availability, probs,
nofly_prob)
# no utility
self.assertTrue('27331928_2490' not in class_level)
# low utility and no observed demand
self.assertEqual(class_level['27334478_3590'], {
'demand': 0,
'spill': 0,
'recapture': 0
})
# one observation and low utility
self.assertLess(class_level['29255538_3590']['demand'],
observed['29255538_3590'])
self.assertLess(class_level['29255538_3590']['spill'], 0.1)
# high utility, many bookings, low availability
self.assertGreater(class_level['29255578_2990']['demand'],
observed['29255578_2990'])
self.assertGreater(class_level['29255578_2990']['spill'],
observed['29255578_2990'])
# moderate utility, many bookings, moderate availability
self.assertGreater(class_level['29255578_3950']['demand'],
observed['29255578_3950'])