Beispiel #1
0
def test_dataframes_mnl5(mtc):
    m5 = NumbaModel()

    from larch.roles import P, X, PX

    m5.utility_co[2] = P("ASC_SR2") * X("1") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") * X("1") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") * X("1") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") * X("1") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") * X("1") + P("hhinc#6") * X("hhinc")
    m5.utility_ca = PX("tottime") + PX("totcost")

    m5.dataframes = mtc

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
    }

    m5.choice_ca_var = 'chose'
    m5.availability_var = '_avail_'
    ll2 = m5.loglike2(beta_in1, return_series=True)

    q1_dll = {
        'ASC_BIKE': -139.43832,
        'ASC_SR2': -788.00574,
        'ASC_SR3P': -126.84879,
        'ASC_TRAN': -357.75186,
        'ASC_WALK': -116.137886,
        'hhinc#2': -46416.28,
        'hhinc#3': -8353.63,
        'hhinc#4': -21409.012,
        'hhinc#5': -8299.654,
        'hhinc#6': -7395.375,
        'totcost': 39520.043,
        'tottime': -26556.303,
    }

    assert ll2.ll == approx(-4930.3212890625)

    for k in q1_dll:
        assert q1_dll[k] == approx(dict(
            ll2.dll)[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict(ll2.dll)[k]}"

    # Test calculate_parameter_covariance doesn't choke if all holdfasts are on:
    m5.lock_values(*beta_in1.keys())
    m5.calculate_parameter_covariance()

    assert np.all(m5.pf['std_err'] == 0)
    assert np.all(m5.pf['robust_std_err'] == 0)
Beispiel #2
0
def test_mtc_with_dataset(mtc_dataset):
    pytest.importorskip("sharrow")
    m = NumbaModel(alts=mtc_dataset['_altid_'].values)
    from larch.roles import P, X, PX
    m.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m.utility_ca = PX("tottime") + PX("totcost")
    m.availability_var = 'avail'
    m.choice_ca_var = 'chose'
    m.datatree = mtc_dataset
    assert m.loglike() == approx(-7309.600971749634)
    m.set_cap(20)
    result = m.maximize_loglike(method='slsqp')
    assert result.loglike == approx(-3626.1862595453385)
Beispiel #3
0
def test_dataframes_mnl5_co(mtc):
    m5 = NumbaModel()

    from larch.roles import P, X, PX
    m5.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m5.dataframes = mtc

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
    }

    ll2 = m5.loglike2(beta_in1, return_series=True)

    q1_dll = {
        'ASC_BIKE': -139.2947,
        'ASC_SR2': -598.531,
        'ASC_SR3P': -77.68647,
        'ASC_TRAN': -715.4206,
        'ASC_WALK': -235.8408,
        'hhinc#2': -35611.855,
        'hhinc#3': -5276.0254,
        'hhinc#4': -42263.88,
        'hhinc#5': -8355.174,
        'hhinc#6': -13866.567,
    }

    assert -5594.70654296875 == approx(ll2.ll, rel=1e-5)

    for k in q1_dll:
        assert q1_dll[k] == approx(dict(
            ll2.dll)[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict(ll2.dll)[k]}"
Beispiel #4
0
def test_constrained_optimization():
    from larch.numba import example
    m = example(1)
    from larch.model.constraints import RatioBound, OrderingBound
    m.set_value("totcost", -0.001, maximum=0)
    m.set_value("tottime", maximum=0)
    m.constraints = [
        RatioBound(P("totcost"),
                   P("tottime"),
                   min_ratio=0.1,
                   max_ratio=1.0,
                   scale=1),
        OrderingBound(P("ASC_WALK"), P("ASC_BIKE")),
    ]
    m.load_data()
    r = m.maximize_loglike(method='slsqp')
    assert r.loglike == approx(-3647.76149525901)
    x = {
        'ASC_BIKE': -0.8087472748965431,
        'ASC_SR2': -2.193449976582375,
        'ASC_SR3P': -3.744188833076006,
        'ASC_TRAN': -0.7603092451373663,
        'ASC_WALK': -0.8087472751682576,
        'hhinc#2': -0.0021699330391421407,
        'hhinc#3': 0.0003696763687090173,
        'hhinc#4': -0.00509836274463602,
        'hhinc#5': -0.0431749907425252,
        'hhinc#6': -0.002373556571769923,
        'totcost': -0.004910169034222911,
        'tottime': -0.04790588175791953,
    }
    assert dict(r.x) == approx(x)
    assert r.iteration_number == 48

    m.set_values("null")
    m.set_value("totcost", -0.001, maximum=0)
    r2 = m.maximize_loglike(method='slsqp', bhhh_start=3)
    assert r2.iteration_number == 24
    assert r2.loglike == approx(-3647.76149525901)
    assert dict(r2.x) == approx(x, rel=1e-2)
Beispiel #5
0
def test_latent_class_mixed_data(swissmetro_raw_df):

    dfs = larch.DataFrames(swissmetro_raw_df, alt_codes=[1, 2, 3])

    m1 = larch.Model(dataservice=dfs)
    m1.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m1.choice_co_code = 'CHOICE'
    m1.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_CO*(GA==0)") * P("B_COST")
    m1.utility_co[2] = X("SM_CO*(GA==0)") * P("B_COST")
    m1.utility_co[3] = P("ASC_CAR") + X("CAR_CO") * P("B_COST")

    m2 = larch.Model(dataservice=dfs)
    m2.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m2.choice_co_code = 'CHOICE'
    m2.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_TT") * P("B_TIME") + X(
        "TRAIN_CO*(GA==0)") * P("B_COST")
    m2.utility_co[
        2] = X("SM_TT") * P("B_TIME") + X("SM_CO*(GA==0)") * P("B_COST")
    m2.utility_co[3] = P(
        "ASC_CAR") + X("CAR_TT") * P("B_TIME") + X("CAR_CO") * P("B_COST")

    dfs2 = larch.DataFrames(swissmetro_raw_df, alt_codes=[1, 2])
    km = larch.Model(dataservice=dfs2)
    km.utility_co[2] = P.W_OTHER

    from larch.model.latentclass import LatentClassModel
    with raises(ValueError):
        m = LatentClassModel(km, {1: m1, 2: m2})
Beispiel #6
0
def test_latent_class(swissmetro_raw_df):

    dfs = larch.DataFrames(swissmetro_raw_df, alt_codes=[1, 2, 3])

    m1 = larch.Model(dataservice=dfs)
    m1.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m1.choice_co_code = 'CHOICE'
    m1.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_CO*(GA==0)") * P("B_COST")
    m1.utility_co[2] = X("SM_CO*(GA==0)") * P("B_COST")
    m1.utility_co[3] = P("ASC_CAR") + X("CAR_CO") * P("B_COST")

    m2 = larch.Model(dataservice=dfs)
    m2.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m2.choice_co_code = 'CHOICE'
    m2.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_TT") * P("B_TIME") + X(
        "TRAIN_CO*(GA==0)") * P("B_COST")
    m2.utility_co[
        2] = X("SM_TT") * P("B_TIME") + X("SM_CO*(GA==0)") * P("B_COST")
    m2.utility_co[3] = P(
        "ASC_CAR") + X("CAR_TT") * P("B_TIME") + X("CAR_CO") * P("B_COST")

    km = larch.Model()
    km.utility_co[2] = P.W_OTHER

    from larch.model.latentclass import LatentClassModel
    m = LatentClassModel(km, {1: m1, 2: m2})

    m.load_data()

    m.set_value(P.ASC_CAR, 0.125 / 2)
    m.set_value(P.ASC_TRAIN, -0.398 / 2)
    m.set_value(P.B_COST, -.0126 / 2)
    m.set_value(P.B_TIME, -0.028 / 2)
    m.set_value(P.W_OTHER, 1.095 / 2)

    check1 = m.check_d_loglike()

    assert dict(check1.data.analytic) == approx({
        'ASC_CAR': -81.69736186616234,
        'ASC_TRAIN': -613.131371089499,
        'B_COST': -6697.31706964777,
        'B_TIME': -40104.940072046316,
        'W_OTHER': 245.43145056623683,
    })

    assert check1.data.similarity.min() > 4

    m.set_value(P.ASC_CAR, 0.125)
    m.set_value(P.ASC_TRAIN, -0.398)
    m.set_value(P.B_COST, -.0126)
    m.set_value(P.B_TIME, -0.028)
    m.set_value(P.W_OTHER, 1.095)

    assert m.loglike() == approx(-5208.502259337974)

    check2 = m.check_d_loglike()

    assert dict(check2.data.analytic) == approx({
        'ASC_CAR': 0.6243716033364302,
        'ASC_TRAIN': 0.9297965389102578,
        'B_COST': -154.03997923797007,
        'B_TIME': 76.19297915128493,
        'W_OTHER': -0.7936963902343083,
    })

    assert check2.data.similarity.min(
    ) > 2  # similarity is a bit lower very close to the optimum
Beispiel #7
0
def test_latent_class_full_data(swissmetro_raw_df):

    dfs = larch.DataFrames(swissmetro_raw_df, alt_codes=[1, 2, 3])

    m1 = larch.Model(dataservice=dfs)
    m1.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m1.choice_co_code = 'CHOICE'
    m1.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_CO*(GA==0)") * P("B_COST")
    m1.utility_co[2] = X("SM_CO*(GA==0)") * P("B_COST")
    m1.utility_co[3] = P("ASC_CAR") + X("CAR_CO") * P("B_COST")

    m2 = larch.Model(dataservice=dfs)
    m2.availability_co_vars = {
        1: "TRAIN_AV_SP",
        2: "SM_AV",
        3: "CAR_AV_SP",
    }
    m2.choice_co_code = 'CHOICE'
    m2.utility_co[1] = P("ASC_TRAIN") + X("TRAIN_TT") * P("B_TIME") + X(
        "TRAIN_CO*(GA==0)") * P("B_COST")
    m2.utility_co[
        2] = X("SM_TT") * P("B_TIME") + X("SM_CO*(GA==0)") * P("B_COST")
    m2.utility_co[3] = P(
        "ASC_CAR") + X("CAR_TT") * P("B_TIME") + X("CAR_CO") * P("B_COST")

    km = larch.Model(dataservice=dfs, alts=[1, 2])
    km.utility_co[2] = P.W_OTHER

    from larch.model.latentclass import LatentClassModel
    m = LatentClassModel(km, {1: m1, 2: m2})

    m.load_data()

    m.set_value(P.ASC_CAR, 0.125 / 2)
    m.set_value(P.ASC_TRAIN, -0.398 / 2)
    m.set_value(P.B_COST, -.0126 / 2)
    m.set_value(P.B_TIME, -0.028 / 2)
    m.set_value(P.W_OTHER, 1.095 / 2)

    check1 = m.check_d_loglike()

    assert dict(check1.data.analytic) == approx({
        'ASC_CAR': -81.69736186616234,
        'ASC_TRAIN': -613.131371089499,
        'B_COST': -6697.31706964777,
        'B_TIME': -40104.940072046316,
        'W_OTHER': 245.43145056623683,
    })
Beispiel #8
0
def test_dataframes_mnl5qt(mtcq):

    m5 = NumbaModel()

    from larch.roles import P, X, PX
    m5.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m5.utility_ca = PX("tottime") + PX("totcost")

    m5.quantity_ca = (+P("FakeSizeAlt") * X('altnum+1') +
                      P("FakeSizeIvtt") * X('ivtt+1'))

    m5.quantity_scale = P.Theta

    m5.dataframes = mtcq

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
        'FakeSizeAlt': 0.123,
        'Theta': 1.0,
    }

    ll2 = m5.loglike2(beta_in1, return_series=True)

    q1_dll = {
        'ASC_BIKE': -272.10342,
        'ASC_SR2': -884.91547,
        'ASC_SR3P': -181.50142,
        'ASC_TRAN': -519.74567,
        'ASC_WALK': -37.595825,
        'FakeSizeAlt': -104.971085,
        'FakeSizeIvtt': 104.971085,
        'hhinc#2': -51884.465,
        'hhinc#3': -11712.436,
        'hhinc#4': -30848.334,
        'hhinc#5': -15970.957,
        'hhinc#6': -3269.796,
        'totcost': 59049.66,
        'tottime': -34646.656,
        'Theta': -838.5296020507812,
    }

    assert ll2.ll == approx(-5598.75244140625, rel=1e-5), f"ll2.ll={ll2.ll}"

    for k in q1_dll:
        assert q1_dll[k] == approx(dict(
            ll2.dll)[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict(ll2.dll)[k]}"

    correct_null_dloglike = {
        'ASC_SR2': -676.598075,
        'ASC_SR3P': -1166.26503,
        'ASC_TRAN': -491.818432,
        'ASC_BIKE': -443.123432,
        'ASC_WALK': 3.970966339111328,
        'FakeSizeAlt': -86.9414603,
        'FakeSizeIvtt': 86.94156646728516,
        'hhinc#2': -40249.0548,
        'hhinc#3': -67312.464,
        'hhinc#4': -30693.2152,
        'hhinc#5': -27236.7637,
        'hhinc#6': -1389.66274,
        'totcost': 145788.421875,
        'tottime': -48732.99609375,
        'Theta': -1362.409129,
    }

    ll0 = m5.loglike2('null', return_series=True)
    assert (ll0.ll == approx(-8486.55377320886))
    dict_ll0_dll = dict(ll0.dll)
    for k in dict_ll0_dll:
        assert dict_ll0_dll[k] == approx(
            correct_null_dloglike[k], rel=1e-5
        ), f'{k}  {dict_ll0_dll[k]} == {(correct_null_dloglike[k])}'
Beispiel #9
0
def test_weighted_nl_bhhh(mtc2):
    j1, j2 = mtc2

    m5 = NumbaModel()

    from larch.roles import P, X, PX
    m5.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m5.utility_ca = PX("tottime") + PX("totcost")

    m5.initialize_graph(alternative_codes=[1, 2, 3, 4, 5, 6])
    m5.graph.add_node(10, children=(5, 6), parameter='MU_nonmotor')
    m5.graph.add_node(11, children=(1, 2, 3), parameter='MU_car')

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
    }

    m5.pf_sort()

    m5.dataframes = j1
    ll1 = m5.loglike2(beta_in1)
    dll1 = m5.d_loglike(beta_in1, return_series=True)
    bhhh1 = m5.bhhh(beta_in1, return_dataframe=True)

    from larch.model import PERSIST_ALL
    ll1 = m5.loglike2_bhhh(beta_in1, return_series=True, persist=PERSIST_ALL)

    m5.dataframes = j2
    m5.mangle()
    ll2 = m5.loglike2_bhhh(beta_in1, return_series=True, persist=PERSIST_ALL)

    q1_dll = {
        'ASC_BIKE': -518.3145850719714,
        'ASC_SR2': 6659.9870966633935,
        'ASC_SR3P': -702.5461471592637,
        'ASC_TRAN': -2069.2556854096474,
        'ASC_WALK': -680.4136747673049,
        'hhinc#2': 390300.04704708763,
        'hhinc#3': -44451.89987844542,
        'hhinc#4': -117769.88300441334,
        'hhinc#5': -29774.93396444093,
        'hhinc#6': -36754.12651709895,
        'totcost': -280658.27799924824,
        'tottime': -66172.15328009706,
    }

    assert (j1.weight_normalization, j2.weight_normalization) == (3.0, 1.5)
    assert (ll1.ll, ll2.ll) == approx(
        (-18829.858031378415, -18829.858031378433))
    dict_ll1_dll = dict(ll1.dll)
    dict_ll2_dll = dict(ll2.dll)
    for k in q1_dll:
        assert q1_dll[k] == approx(
            dict_ll1_dll[k], rel=1e-5), f"{k} {q1_dll[k]} = {dict_ll1_dll[k]}"
        assert q1_dll[k] == approx(
            dict_ll2_dll[k], rel=1e-5), f"{k} {q1_dll[k]} = {dict_ll2_dll[k]}"

    bhhh_correct = {
        ('ASC_BIKE', 'ASC_BIKE'): 102.45820678523614,
        ('ASC_BIKE', 'ASC_SR2'): -285.118579955482,
        ('ASC_BIKE', 'ASC_SR3P'): 19.760852749005203,
        ('ASC_BIKE', 'ASC_TRAN'): 76.573987238185,
        ('ASC_BIKE', 'ASC_WALK'): 32.348453110164314,
        ('ASC_BIKE', 'MU_car'): -219.27982884201612,
        ('ASC_BIKE', 'MU_nonmotor'): 69.38016083689357,
        ('ASC_BIKE', 'hhinc#2'): -16063.872652571805,
        ('ASC_BIKE', 'hhinc#3'): 1231.2989480753972,
        ('ASC_BIKE', 'hhinc#4'): 4436.148875125189,
        ('ASC_BIKE', 'hhinc#5'): 5322.601183863066,
        ('ASC_BIKE', 'hhinc#6'): 1870.2419763938155,
        ('ASC_BIKE', 'totcost'): 651.7010091466225,
        ('ASC_BIKE', 'tottime'): 3410.093422258546,
        ('ASC_SR2', 'ASC_BIKE'): -285.118579955482,
        ('ASC_SR2', 'ASC_SR2'): 6156.860848428152,
        ('ASC_SR2', 'ASC_SR3P'): -412.00155512537197,
        ('ASC_SR2', 'ASC_TRAN'): -1312.7312255613883,
        ('ASC_SR2', 'ASC_WALK'): -461.36148965146094,
        ('ASC_SR2', 'MU_car'): 3491.370489213462,
        ('ASC_SR2', 'MU_nonmotor'): -244.85789490150424,
        ('ASC_SR2', 'hhinc#2'): 360960.5064929566,
        ('ASC_SR2', 'hhinc#3'): -25836.407171358624,
        ('ASC_SR2', 'hhinc#4'): -73800.95798232155,
        ('ASC_SR2', 'hhinc#5'): -16063.872652571805,
        ('ASC_SR2', 'hhinc#6'): -23821.208247582188,
        ('ASC_SR2', 'totcost'): -255314.7240133045,
        ('ASC_SR2', 'tottime'): -26043.721750561366,
        ('ASC_SR3P', 'ASC_BIKE'): 19.760852749005203,
        ('ASC_SR3P', 'ASC_SR2'): -412.00155512537197,
        ('ASC_SR3P', 'ASC_SR3P'): 194.01812799791654,
        ('ASC_SR3P', 'ASC_TRAN'): 75.72048947864153,
        ('ASC_SR3P', 'ASC_WALK'): 21.26889182584585,
        ('ASC_SR3P', 'MU_car'): 108.36680636408329,
        ('ASC_SR3P', 'MU_nonmotor'): 11.214695666269147,
        ('ASC_SR3P', 'hhinc#2'): -25836.407171358624,
        ('ASC_SR3P', 'hhinc#3'): 11951.689201021742,
        ('ASC_SR3P', 'hhinc#4'): 4654.093182472214,
        ('ASC_SR3P', 'hhinc#5'): 1231.298948075397,
        ('ASC_SR3P', 'hhinc#6'): 1291.5421535124285,
        ('ASC_SR3P', 'totcost'): -873.8405986324798,
        ('ASC_SR3P', 'tottime'): 2847.414853230762,
        ('ASC_TRAN', 'ASC_BIKE'): 76.573987238185,
        ('ASC_TRAN', 'ASC_SR2'): -1312.7312255613883,
        ('ASC_TRAN', 'ASC_SR3P'): 75.72048947864153,
        ('ASC_TRAN', 'ASC_TRAN'): 795.8802149505714,
        ('ASC_TRAN', 'ASC_WALK'): 106.98212315599287,
        ('ASC_TRAN', 'MU_car'): -930.0194408568389,
        ('ASC_TRAN', 'MU_nonmotor'): 58.620406514681704,
        ('ASC_TRAN', 'hhinc#2'): -73800.95798232155,
        ('ASC_TRAN', 'hhinc#3'): 4654.093182472214,
        ('ASC_TRAN', 'hhinc#4'): 43744.97724907036,
        ('ASC_TRAN', 'hhinc#5'): 4436.148875125189,
        ('ASC_TRAN', 'hhinc#6'): 5766.003794801971,
        ('ASC_TRAN', 'totcost'): -289.26636202646205,
        ('ASC_TRAN', 'tottime'): 17843.995404571957,
        ('ASC_WALK', 'ASC_BIKE'): 32.348453110164314,
        ('ASC_WALK', 'ASC_SR2'): -461.36148965146094,
        ('ASC_WALK', 'ASC_SR3P'): 21.26889182584585,
        ('ASC_WALK', 'ASC_TRAN'): 106.98212315599287,
        ('ASC_WALK', 'ASC_WALK'): 257.4268216790701,
        ('ASC_WALK', 'MU_car'): -372.1302408319178,
        ('ASC_WALK', 'MU_nonmotor'): 84.61754029339109,
        ('ASC_WALK', 'hhinc#2'): -23821.208247582188,
        ('ASC_WALK', 'hhinc#3'): 1291.5421535124283,
        ('ASC_WALK', 'hhinc#4'): 5766.003794801973,
        ('ASC_WALK', 'hhinc#5'): 1870.2419763938155,
        ('ASC_WALK', 'hhinc#6'): 12476.241347103962,
        ('ASC_WALK', 'totcost'): -4472.049603002317,
        ('ASC_WALK', 'tottime'): 7147.124392574635,
        ('MU_car', 'ASC_BIKE'): -219.27982884201612,
        ('MU_car', 'ASC_SR2'): 3491.370489213462,
        ('MU_car', 'ASC_SR3P'): 108.36680636408329,
        ('MU_car', 'ASC_TRAN'): -930.0194408568389,
        ('MU_car', 'ASC_WALK'): -372.1302408319178,
        ('MU_car', 'MU_car'): 3048.745424759559,
        ('MU_car', 'MU_nonmotor'): -215.09078801898488,
        ('MU_car', 'hhinc#2'): 209343.58472860648,
        ('MU_car', 'hhinc#3'): 4904.0435782757395,
        ('MU_car', 'hhinc#4'): -54645.9131974324,
        ('MU_car', 'hhinc#5'): -12553.188578813708,
        ('MU_car', 'hhinc#6'): -20397.013767114706,
        ('MU_car', 'totcost'): -130926.84443581512,
        ('MU_car', 'tottime'): -19903.25282015786,
        ('MU_nonmotor', 'ASC_BIKE'): 69.38016083689357,
        ('MU_nonmotor', 'ASC_SR2'): -244.85789490150424,
        ('MU_nonmotor', 'ASC_SR3P'): 11.214695666269147,
        ('MU_nonmotor', 'ASC_TRAN'): 58.620406514681704,
        ('MU_nonmotor', 'ASC_WALK'): 84.61754029339109,
        ('MU_nonmotor', 'MU_car'): -215.09078801898488,
        ('MU_nonmotor', 'MU_nonmotor'): 106.11047953795583,
        ('MU_nonmotor', 'hhinc#2'): -13788.1097937347,
        ('MU_nonmotor', 'hhinc#3'): 711.5550773867994,
        ('MU_nonmotor', 'hhinc#4'): 3435.2281478513137,
        ('MU_nonmotor', 'hhinc#5'): 3589.230542100178,
        ('MU_nonmotor', 'hhinc#6'): 4773.341396211997,
        ('MU_nonmotor', 'totcost'): -571.9687206358277,
        ('MU_nonmotor', 'tottime'): 3082.561197668304,
        ('hhinc#2', 'ASC_BIKE'): -16063.872652571805,
        ('hhinc#2', 'ASC_SR2'): 360960.5064929566,
        ('hhinc#2', 'ASC_SR3P'): -25836.407171358624,
        ('hhinc#2', 'ASC_TRAN'): -73800.95798232155,
        ('hhinc#2', 'ASC_WALK'): -23821.208247582188,
        ('hhinc#2', 'MU_car'): 209343.58472860648,
        ('hhinc#2', 'MU_nonmotor'): -13788.1097937347,
        ('hhinc#2', 'hhinc#2'): 27739015.863625906,
        ('hhinc#2', 'hhinc#3'): -2107887.124567991,
        ('hhinc#2', 'hhinc#4'): -5551797.986970259,
        ('hhinc#2', 'hhinc#5'): -1180261.9541857818,
        ('hhinc#2', 'hhinc#6'): -1731206.5786703678,
        ('hhinc#2', 'totcost'): -15915701.570008647,
        ('hhinc#2', 'tottime'): -1404099.9397647784,
        ('hhinc#3', 'ASC_BIKE'): 1231.2989480753972,
        ('hhinc#3', 'ASC_SR2'): -25836.407171358624,
        ('hhinc#3', 'ASC_SR3P'): 11951.689201021742,
        ('hhinc#3', 'ASC_TRAN'): 4654.093182472214,
        ('hhinc#3', 'ASC_WALK'): 1291.5421535124283,
        ('hhinc#3', 'MU_car'): 4904.0435782757395,
        ('hhinc#3', 'MU_nonmotor'): 711.5550773867994,
        ('hhinc#3', 'hhinc#2'): -2107887.124567991,
        ('hhinc#3', 'hhinc#3'): 985365.3310746389,
        ('hhinc#3', 'hhinc#4'): 365082.18492341647,
        ('hhinc#3', 'hhinc#5'): 96863.55545307972,
        ('hhinc#3', 'hhinc#6'): 105646.64587551646,
        ('hhinc#3', 'totcost'): -18197.96776085889,
        ('hhinc#3', 'tottime'): 173393.88742844868,
        ('hhinc#4', 'ASC_BIKE'): 4436.148875125189,
        ('hhinc#4', 'ASC_SR2'): -73800.95798232155,
        ('hhinc#4', 'ASC_SR3P'): 4654.093182472214,
        ('hhinc#4', 'ASC_TRAN'): 43744.97724907036,
        ('hhinc#4', 'ASC_WALK'): 5766.003794801973,
        ('hhinc#4', 'MU_car'): -54645.9131974324,
        ('hhinc#4', 'MU_nonmotor'): 3435.2281478513137,
        ('hhinc#4', 'hhinc#2'): -5551797.986970259,
        ('hhinc#4', 'hhinc#3'): 365082.18492341647,
        ('hhinc#4', 'hhinc#4'): 3238425.3752979534,
        ('hhinc#4', 'hhinc#5'): 328415.284903046,
        ('hhinc#4', 'hhinc#6'): 431510.21040576114,
        ('hhinc#4', 'totcost'): -203942.9008935652,
        ('hhinc#4', 'tottime'): 996131.649172921,
        ('hhinc#5', 'ASC_BIKE'): 5322.601183863066,
        ('hhinc#5', 'ASC_SR2'): -16063.872652571805,
        ('hhinc#5', 'ASC_SR3P'): 1231.298948075397,
        ('hhinc#5', 'ASC_TRAN'): 4436.148875125189,
        ('hhinc#5', 'ASC_WALK'): 1870.2419763938155,
        ('hhinc#5', 'MU_car'): -12553.188578813708,
        ('hhinc#5', 'MU_nonmotor'): 3589.230542100178,
        ('hhinc#5', 'hhinc#2'): -1180261.9541857818,
        ('hhinc#5', 'hhinc#3'): 96863.55545307972,
        ('hhinc#5', 'hhinc#4'): 328415.284903046,
        ('hhinc#5', 'hhinc#5'): 376816.62222404545,
        ('hhinc#5', 'hhinc#6'): 139543.8152706944,
        ('hhinc#5', 'totcost'): 64871.45953365474,
        ('hhinc#5', 'tottime'): 191184.20323081187,
        ('hhinc#6', 'ASC_BIKE'): 1870.2419763938155,
        ('hhinc#6', 'ASC_SR2'): -23821.208247582188,
        ('hhinc#6', 'ASC_SR3P'): 1291.5421535124285,
        ('hhinc#6', 'ASC_TRAN'): 5766.003794801971,
        ('hhinc#6', 'ASC_WALK'): 12476.241347103962,
        ('hhinc#6', 'MU_car'): -20397.013767114706,
        ('hhinc#6', 'MU_nonmotor'): 4773.341396211997,
        ('hhinc#6', 'hhinc#2'): -1731206.5786703678,
        ('hhinc#6', 'hhinc#3'): 105646.64587551646,
        ('hhinc#6', 'hhinc#4'): 431510.21040576114,
        ('hhinc#6', 'hhinc#5'): 139543.8152706944,
        ('hhinc#6', 'hhinc#6'): 872604.6192807175,
        ('hhinc#6', 'totcost'): -95081.8079551341,
        ('hhinc#6', 'tottime'): 364196.75026433234,
        ('totcost', 'ASC_BIKE'): 651.7010091466225,
        ('totcost', 'ASC_SR2'): -255314.7240133045,
        ('totcost', 'ASC_SR3P'): -873.8405986324798,
        ('totcost', 'ASC_TRAN'): -289.26636202646205,
        ('totcost', 'ASC_WALK'): -4472.049603002317,
        ('totcost', 'MU_car'): -130926.84443581512,
        ('totcost', 'MU_nonmotor'): -571.9687206358277,
        ('totcost', 'hhinc#2'): -15915701.570008647,
        ('totcost', 'hhinc#3'): -18197.96776085889,
        ('totcost', 'hhinc#4'): -203942.9008935652,
        ('totcost', 'hhinc#5'): 64871.45953365474,
        ('totcost', 'hhinc#6'): -95081.8079551341,
        ('totcost', 'totcost'): 67516567.00440338,
        ('totcost', 'tottime'): -775245.3645765011,
        ('tottime', 'ASC_BIKE'): 3410.093422258546,
        ('tottime', 'ASC_SR2'): -26043.721750561366,
        ('tottime', 'ASC_SR3P'): 2847.414853230762,
        ('tottime', 'ASC_TRAN'): 17843.995404571957,
        ('tottime', 'ASC_WALK'): 7147.124392574635,
        ('tottime', 'MU_car'): -19903.25282015786,
        ('tottime', 'MU_nonmotor'): 3082.561197668304,
        ('tottime', 'hhinc#2'): -1404099.9397647784,
        ('tottime', 'hhinc#3'): 173393.88742844868,
        ('tottime', 'hhinc#4'): 996131.649172921,
        ('tottime', 'hhinc#5'): 191184.20323081187,
        ('tottime', 'hhinc#6'): 364196.75026433234,
        ('tottime', 'totcost'): -775245.3645765011,
        ('tottime', 'tottime'): 910724.9012464393,
    }
    assert dict(ll1.bhhh.unstack()) == approx(bhhh_correct)
    assert dict(ll2.bhhh.unstack()) == approx(bhhh_correct)

    dll_casewise_A = ll2.dll_casewise / j2.weight_normalization
    dll_casewise_B = np.asarray(dll_casewise_A) / j2.data_wt.values

    corrected_bhhh = pd.DataFrame(
        np.dot(dll_casewise_A.T, dll_casewise_B),
        index=ll2.dll.index,
        columns=ll2.dll.index,
    ) * j2.weight_normalization

    assert dict(ll1.bhhh.unstack()) == approx(dict(corrected_bhhh.unstack()))
Beispiel #10
0
def test_weighted_bhhh(mtc2):
    j1, j2 = mtc2

    m5 = NumbaModel()

    from larch.roles import P, X, PX
    m5.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m5.utility_ca = PX("tottime") + PX("totcost")

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
    }

    m5.dataframes = j1
    m5.pf_sort()
    ll1 = m5.loglike(beta_in1)
    dll1 = m5.d_loglike(beta_in1, return_series=True)
    bhhh1 = m5.bhhh(beta_in1, return_dataframe=True)

    m5.dataframes = j2
    m5.pf_sort()
    ll2 = m5.loglike(beta_in1)
    dll2 = m5.d_loglike(beta_in1, return_series=True)
    bhhh2 = m5.bhhh(beta_in1, return_dataframe=True)

    q1_dll = {
        'ASC_BIKE': -518.3145850719714,
        'ASC_SR2': 6659.9870966633935,
        'ASC_SR3P': -702.5461471592637,
        'ASC_TRAN': -2069.2556854096474,
        'ASC_WALK': -680.4136747673049,
        'hhinc#2': 390300.04704708763,
        'hhinc#3': -44451.89987844542,
        'hhinc#4': -117769.88300441334,
        'hhinc#5': -29774.93396444093,
        'hhinc#6': -36754.12651709895,
        'totcost': -280658.27799924824,
        'tottime': -66172.15328009706,
    }

    assert (j1.weight_normalization, j2.weight_normalization) == (3.0, 1.5)

    assert (ll1, ll2) == approx((-18829.858031378415, -18829.858031378433))
    dict_ll1_dll = dict(dll1)
    dict_ll2_dll = dict(dll2)

    for k in q1_dll:
        assert q1_dll[k] == approx(
            dict_ll1_dll[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict_ll1_dll[k]}"
        assert q1_dll[k] == approx(
            dict_ll2_dll[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict_ll2_dll[k]}"

    bhhh_correct = {
        ('ASC_BIKE', 'ASC_BIKE'): 102.45820678523617,
        ('ASC_BIKE', 'ASC_SR2'): -285.118579955482,
        ('ASC_BIKE', 'ASC_SR3P'): 19.760852749005203,
        ('ASC_BIKE', 'ASC_TRAN'): 76.57398723818505,
        ('ASC_BIKE', 'ASC_WALK'): 32.34845311016433,
        ('ASC_BIKE', 'hhinc#2'): -16063.872652571783,
        ('ASC_BIKE', 'hhinc#3'): 1231.2989480753972,
        ('ASC_BIKE', 'hhinc#4'): 4436.148875125188,
        ('ASC_BIKE', 'hhinc#5'): 5322.601183863066,
        ('ASC_BIKE', 'hhinc#6'): 1870.2419763938155,
        ('ASC_BIKE', 'totcost'): 651.7010091466211,
        ('ASC_BIKE', 'tottime'): 3410.0934222585456,
        ('ASC_SR2', 'ASC_BIKE'): -285.118579955482,
        ('ASC_SR2', 'ASC_SR2'): 6156.860848428149,
        ('ASC_SR2', 'ASC_SR3P'): -412.00155512537174,
        ('ASC_SR2', 'ASC_TRAN'): -1312.7312255613908,
        ('ASC_SR2', 'ASC_WALK'): -461.36148965146094,
        ('ASC_SR2', 'hhinc#2'): 360960.5064929568,
        ('ASC_SR2', 'hhinc#3'): -25836.407171358598,
        ('ASC_SR2', 'hhinc#4'): -73800.95798232155,
        ('ASC_SR2', 'hhinc#5'): -16063.872652571783,
        ('ASC_SR2', 'hhinc#6'): -23821.20824758219,
        ('ASC_SR2', 'totcost'): -255314.72401330443,
        ('ASC_SR2', 'tottime'): -26043.72175056138,
        ('ASC_SR3P', 'ASC_BIKE'): 19.760852749005203,
        ('ASC_SR3P', 'ASC_SR2'): -412.00155512537174,
        ('ASC_SR3P', 'ASC_SR3P'): 194.01812799791645,
        ('ASC_SR3P', 'ASC_TRAN'): 75.72048947864153,
        ('ASC_SR3P', 'ASC_WALK'): 21.268891825845834,
        ('ASC_SR3P', 'hhinc#2'): -25836.407171358598,
        ('ASC_SR3P', 'hhinc#3'): 11951.689201021733,
        ('ASC_SR3P', 'hhinc#4'): 4654.093182472213,
        ('ASC_SR3P', 'hhinc#5'): 1231.2989480753972,
        ('ASC_SR3P', 'hhinc#6'): 1291.5421535124271,
        ('ASC_SR3P', 'totcost'): -873.8405986324893,
        ('ASC_SR3P', 'tottime'): 2847.414853230761,
        ('ASC_TRAN', 'ASC_BIKE'): 76.57398723818505,
        ('ASC_TRAN', 'ASC_SR2'): -1312.7312255613908,
        ('ASC_TRAN', 'ASC_SR3P'): 75.72048947864153,
        ('ASC_TRAN', 'ASC_TRAN'): 795.8802149505715,
        ('ASC_TRAN', 'ASC_WALK'): 106.98212315599287,
        ('ASC_TRAN', 'hhinc#2'): -73800.95798232155,
        ('ASC_TRAN', 'hhinc#3'): 4654.093182472214,
        ('ASC_TRAN', 'hhinc#4'): 43744.97724907039,
        ('ASC_TRAN', 'hhinc#5'): 4436.148875125188,
        ('ASC_TRAN', 'hhinc#6'): 5766.003794801968,
        ('ASC_TRAN', 'totcost'): -289.2663620264443,
        ('ASC_TRAN', 'tottime'): 17843.995404571946,
        ('ASC_WALK', 'ASC_BIKE'): 32.34845311016433,
        ('ASC_WALK', 'ASC_SR2'): -461.36148965146094,
        ('ASC_WALK', 'ASC_SR3P'): 21.268891825845834,
        ('ASC_WALK', 'ASC_TRAN'): 106.98212315599287,
        ('ASC_WALK', 'ASC_WALK'): 257.42682167907014,
        ('ASC_WALK', 'hhinc#2'): -23821.20824758219,
        ('ASC_WALK', 'hhinc#3'): 1291.5421535124271,
        ('ASC_WALK', 'hhinc#4'): 5766.003794801967,
        ('ASC_WALK', 'hhinc#5'): 1870.2419763938155,
        ('ASC_WALK', 'hhinc#6'): 12476.241347103956,
        ('ASC_WALK', 'totcost'): -4472.049603002317,
        ('ASC_WALK', 'tottime'): 7147.124392574634,
        ('hhinc#2', 'ASC_BIKE'): -16063.872652571783,
        ('hhinc#2', 'ASC_SR2'): 360960.5064929568,
        ('hhinc#2', 'ASC_SR3P'): -25836.407171358598,
        ('hhinc#2', 'ASC_TRAN'): -73800.95798232155,
        ('hhinc#2', 'ASC_WALK'): -23821.20824758219,
        ('hhinc#2', 'hhinc#2'): 27739015.863625936,
        ('hhinc#2', 'hhinc#3'): -2107887.1245679897,
        ('hhinc#2', 'hhinc#4'): -5551797.986970257,
        ('hhinc#2', 'hhinc#5'): -1180261.954185782,
        ('hhinc#2', 'hhinc#6'): -1731206.5786703676,
        ('hhinc#2', 'totcost'): -15915701.570008647,
        ('hhinc#2', 'tottime'): -1404099.9397647786,
        ('hhinc#3', 'ASC_BIKE'): 1231.2989480753972,
        ('hhinc#3', 'ASC_SR2'): -25836.407171358598,
        ('hhinc#3', 'ASC_SR3P'): 11951.689201021733,
        ('hhinc#3', 'ASC_TRAN'): 4654.093182472214,
        ('hhinc#3', 'ASC_WALK'): 1291.5421535124271,
        ('hhinc#3', 'hhinc#2'): -2107887.1245679897,
        ('hhinc#3', 'hhinc#3'): 985365.3310746389,
        ('hhinc#3', 'hhinc#4'): 365082.18492341647,
        ('hhinc#3', 'hhinc#5'): 96863.5554530797,
        ('hhinc#3', 'hhinc#6'): 105646.64587551638,
        ('hhinc#3', 'totcost'): -18197.967760858926,
        ('hhinc#3', 'tottime'): 173393.88742844874,
        ('hhinc#4', 'ASC_BIKE'): 4436.148875125188,
        ('hhinc#4', 'ASC_SR2'): -73800.95798232155,
        ('hhinc#4', 'ASC_SR3P'): 4654.093182472213,
        ('hhinc#4', 'ASC_TRAN'): 43744.97724907039,
        ('hhinc#4', 'ASC_WALK'): 5766.003794801967,
        ('hhinc#4', 'hhinc#2'): -5551797.986970257,
        ('hhinc#4', 'hhinc#3'): 365082.18492341647,
        ('hhinc#4', 'hhinc#4'): 3238425.3752979557,
        ('hhinc#4', 'hhinc#5'): 328415.2849030458,
        ('hhinc#4', 'hhinc#6'): 431510.210405761,
        ('hhinc#4', 'totcost'): -203942.90089356547,
        ('hhinc#4', 'tottime'): 996131.6491729214,
        ('hhinc#5', 'ASC_BIKE'): 5322.601183863066,
        ('hhinc#5', 'ASC_SR2'): -16063.872652571783,
        ('hhinc#5', 'ASC_SR3P'): 1231.2989480753972,
        ('hhinc#5', 'ASC_TRAN'): 4436.148875125188,
        ('hhinc#5', 'ASC_WALK'): 1870.2419763938155,
        ('hhinc#5', 'hhinc#2'): -1180261.954185782,
        ('hhinc#5', 'hhinc#3'): 96863.5554530797,
        ('hhinc#5', 'hhinc#4'): 328415.2849030458,
        ('hhinc#5', 'hhinc#5'): 376816.6222240454,
        ('hhinc#5', 'hhinc#6'): 139543.81527069444,
        ('hhinc#5', 'totcost'): 64871.45953365492,
        ('hhinc#5', 'tottime'): 191184.20323081187,
        ('hhinc#6', 'ASC_BIKE'): 1870.2419763938155,
        ('hhinc#6', 'ASC_SR2'): -23821.20824758219,
        ('hhinc#6', 'ASC_SR3P'): 1291.5421535124271,
        ('hhinc#6', 'ASC_TRAN'): 5766.003794801968,
        ('hhinc#6', 'ASC_WALK'): 12476.241347103956,
        ('hhinc#6', 'hhinc#2'): -1731206.5786703676,
        ('hhinc#6', 'hhinc#3'): 105646.64587551638,
        ('hhinc#6', 'hhinc#4'): 431510.210405761,
        ('hhinc#6', 'hhinc#5'): 139543.81527069444,
        ('hhinc#6', 'hhinc#6'): 872604.6192807176,
        ('hhinc#6', 'totcost'): -95081.80795513398,
        ('hhinc#6', 'tottime'): 364196.75026433205,
        ('totcost', 'ASC_BIKE'): 651.7010091466211,
        ('totcost', 'ASC_SR2'): -255314.72401330443,
        ('totcost', 'ASC_SR3P'): -873.8405986324893,
        ('totcost', 'ASC_TRAN'): -289.2663620264443,
        ('totcost', 'ASC_WALK'): -4472.049603002317,
        ('totcost', 'hhinc#2'): -15915701.570008647,
        ('totcost', 'hhinc#3'): -18197.967760858926,
        ('totcost', 'hhinc#4'): -203942.90089356547,
        ('totcost', 'hhinc#5'): 64871.45953365492,
        ('totcost', 'hhinc#6'): -95081.80795513398,
        ('totcost', 'totcost'): 67516567.00440338,
        ('totcost', 'tottime'): -775245.3645765022,
        ('tottime', 'ASC_BIKE'): 3410.0934222585456,
        ('tottime', 'ASC_SR2'): -26043.72175056138,
        ('tottime', 'ASC_SR3P'): 2847.414853230761,
        ('tottime', 'ASC_TRAN'): 17843.995404571946,
        ('tottime', 'ASC_WALK'): 7147.124392574634,
        ('tottime', 'hhinc#2'): -1404099.9397647786,
        ('tottime', 'hhinc#3'): 173393.88742844874,
        ('tottime', 'hhinc#4'): 996131.6491729214,
        ('tottime', 'hhinc#5'): 191184.20323081187,
        ('tottime', 'hhinc#6'): 364196.75026433205,
        ('tottime', 'totcost'): -775245.3645765022,
        ('tottime', 'tottime'): 910724.9012464394,
    }

    assert dict(bhhh1.unstack()) == approx(bhhh_correct)
    assert dict(bhhh2.unstack()) == approx(bhhh_correct)

    assert m5.check_d_loglike().data.similarity.min() > 4
Beispiel #11
0
def test_dataframes_nl5(mtc):
    m5 = NumbaModel()

    m5.utility_co[2] = P("ASC_SR2") + P("hhinc#2") * X("hhinc")
    m5.utility_co[3] = P("ASC_SR3P") + P("hhinc#3") * X("hhinc")
    m5.utility_co[4] = P("ASC_TRAN") + P("hhinc#4") * X("hhinc")
    m5.utility_co[5] = P("ASC_BIKE") + P("hhinc#5") * X("hhinc")
    m5.utility_co[6] = P("ASC_WALK") + P("hhinc#6") * X("hhinc")
    m5.utility_ca = PX("tottime") + PX("totcost")

    m5.dataframes = mtc

    m5.graph.add_node(9, children=(5, 6), parameter='MU_NonMotorized')

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
    }

    ll2 = m5.loglike2(beta_in1, return_series=True)

    q1_dll = {
        'ASC_BIKE': -139.43832,
        'ASC_SR2': -788.00574,
        'ASC_SR3P': -126.84879,
        'ASC_TRAN': -357.75186,
        'ASC_WALK': -116.137886,
        'hhinc#2': -46416.28,
        'hhinc#3': -8353.63,
        'hhinc#4': -21409.012,
        'hhinc#5': -8299.654,
        'hhinc#6': -7395.375,
        'totcost': 39520.043,
        'tottime': -26556.303,
    }

    assert approx(ll2.ll) == -4930.3212890625
    dict_ll2_dll = dict(ll2.dll)

    for k in q1_dll:
        assert q1_dll[k] == approx(
            dict_ll2_dll[k], rel=1e-5), f"{k} {q1_dll[k]} != {dict_ll2_dll[k]}"

    beta_in1 = {
        'ASC_BIKE': -0.8523646111088327,
        'ASC_SR2': -0.5233769323949348,
        'ASC_SR3P': -2.3202089848081027,
        'ASC_TRAN': -0.05615933557609158,
        'ASC_WALK': 0.050082767550586924,
        'hhinc#2': -0.001040241396513087,
        'hhinc#3': 0.0031822969445656542,
        'hhinc#4': -0.0017162484345735326,
        'hhinc#5': -0.004071521055900851,
        'hhinc#6': -0.0021316332241034445,
        'totcost': -0.001336661560553717,
        'tottime': -0.01862990704919887,
        'MU_NonMotorized': 0.5,
    }

    ll2b = m5.loglike2(beta_in1, return_series=True)

    q1_dllb = {
        'ASC_BIKE': -94.071343,
        'ASC_SR2': -800.092341,
        'ASC_SR3P': -129.354567,
        'ASC_TRAN': -369.808551,
        'ASC_WALK': -114.786728,
        'MU_NonMotorized': -34.816070,
        'hhinc#2': -47089.611079,
        'hhinc#3': -8505.116916,
        'hhinc#4': -22071.859018,
        'hhinc#5': -5844.969336,
        'hhinc#6': -7168.859044,
        'totcost': 37322.528282,
        'tottime': -26479.290942,
    }

    assert -4897.764630665653 == approx(ll2b.ll)

    dict_ll2b_dll = dict(ll2b.dll)

    for k in q1_dllb:
        assert q1_dllb[k] == approx(
            dict_ll2b_dll[k],
            rel=1e-5), f"{k} {q1_dllb[k]} != {dict_ll2b_dll[k]}"

    print(m5.check_d_loglike().data['similarity'].min())
    chk = m5.check_d_loglike()

    assert chk.data['similarity'].min() > 4
Beispiel #12
0
def qmnl_straw_man_model_1():

    from larch.roles import P, X

    altcodes = (1, 2, 3, 4, 5, 6)
    from larch.data_services.examples import MTC
    dt = MTC()

    p = Model(parameters=[], alts=altcodes, dataservice=dt, graph=None)

    from larch.roles import P, X

    p.utility_ca = (+P('tottime') * X('tottime') + P('totcost') * X('totcost'))

    p.utility_co = {
        2: (P('ASC#2') * X('1') + P('hhinc#2') * X('hhinc')),
        3: (P('ASC#3') * X('1') + P('hhinc#3') * X('hhinc')),
        4: (P('ASC#4') * X('1') + P('hhinc#4') * X('hhinc')),
        5: (P('ASC#5') * X('1') + P('hhinc#5') * X('hhinc')),
        6: (P('ASC#6') * X('1') + P('hhinc#6') * X('hhinc')),
    }

    p.quantity_ca = (+P("FakeSizeAlt") * X('altnum+1') +
                     P("FakeSizeIvtt") * X('ivtt+1'))

    p.availability_var = '_avail_'
    p.choice_ca_var = '_choice_'

    p.load_data()
    return p
Beispiel #13
0
def test_eville_mode_with_dataset():
    pytest.importorskip("sharrow")
    from larch.examples import EXAMPVILLE
    tree = EXAMPVILLE('datatree')
    DA = 1
    SR = 2
    Walk = 3
    Bike = 4
    Transit = 5
    m = NumbaModel(
        alts={
            DA: 'DA',
            SR: 'SR',
            Walk: 'Walk',
            Bike: 'Bike',
            Transit: 'Transit',
        },
        datatree=tree,
    )
    m.title = "Exampville Work Tour Mode Choice v1"
    m.utility_co[DA] = (
        +P.InVehTime * X("od.AUTO_TIME + do.AUTO_TIME") +
        P.Cost * X("od.AUTO_COST + do.AUTO_COST")  # dollars per mile
    )
    m.utility_co[SR] = (
        +P.ASC_SR + P.InVehTime * X("od.AUTO_TIME + do.AUTO_TIME") + P.Cost *
        X("od.AUTO_COST + do.AUTO_COST") * 0.5  # dollars per mile, half share
        + P("LogIncome:SR") * X("log(INCOME)"))
    m.utility_co[Walk] = (+P.ASC_Walk +
                          P.NonMotorTime * X("od.WALK_TIME + do.WALK_TIME") +
                          P("LogIncome:Walk") * X("log(INCOME)"))
    m.utility_co[Bike] = (+P.ASC_Bike +
                          P.NonMotorTime * X("od.BIKE_TIME + do.BIKE_TIME") +
                          P("LogIncome:Bike") * X("log(INCOME)"))
    m.utility_co[Transit] = (
        +P.ASC_Transit + P.InVehTime * X("od.TRANSIT_IVTT + do.TRANSIT_IVTT") +
        P.OutVehTime * X("od.TRANSIT_OVTT + do.TRANSIT_OVTT") +
        P.Cost * X("od.TRANSIT_FARE + do.TRANSIT_FARE") +
        P("LogIncome:Transit") * X('log(INCOME)'))
    Car = m.graph.new_node(parameter='Mu:Car', children=[DA, SR], name='Car')
    NonMotor = m.graph.new_node(parameter='Mu:NonMotor',
                                children=[Walk, Bike],
                                name='NonMotor')
    Motor = m.graph.new_node(parameter='Mu:Motor',
                             children=[Car, Transit],
                             name='Motor')
    m.choice_co_code = 'TOURMODE'
    m.availability_co_vars = {
        DA: 'AGE >= 16',
        SR: '1',
        Walk: 'WALK_TIME < 60',
        Bike: 'BIKE_TIME < 60',
        Transit: 'TRANSIT_FARE>0',
    }
    assert m.loglike() == approx(-28846.81581153095)
    assert m.dataflows.keys() == {'co', 'avail_co'}
    assert m.d_loglike() == approx(
        np.array([
            -5.02935000e+03, -2.69235000e+03, -1.72110000e+03, -2.21118333e+03,
            1.67050996e+04, 1.26952101e+05, -5.50687172e+04, -3.03043210e+04,
            -1.93304978e+04, -2.41203790e+04, 5.82518580e+03, 2.92870835e+02,
            -3.31973600e+03, -3.47182502e+05, -2.26234434e+05
        ]))
    m.set_cap(30)
    r = m.maximize_loglike(method='slsqp')
    assert dict(r.x) == approx(
        {
            'ASC_Bike': 1.0207805052947376,
            'ASC_SR': 2.9895850624870013,
            'ASC_Transit': 8.508746668124973,
            'ASC_Walk': 7.473491883981414,
            'Cost': -0.17750345102327197,
            'InVehTime': -0.06760741744874971,
            'LogIncome:Bike': -0.3648653520247778,
            'LogIncome:SR': -0.4222047281799176,
            'LogIncome:Transit': -0.6960684260831979,
            'LogIncome:Walk': -0.4548368480289394,
            'Mu:Car': 0.5466601043649147,
            'Mu:Motor': 0.9540629350784797,
            'Mu:NonMotor': 0.8636508452469879,
            'NonMotorTime': -0.1268142189962758,
            'OutVehTime': -0.14752266135040631,
        },
        rel=1e-5)
    assert r.loglike == approx(-8047.006193851376)
    worktours = tree.query_cases('TOURPURP==1')
    m.datatree = worktours
    assert m.loglike() == approx(-3527.6797690247113)
    m.float_dtype = np.float32
    assert m.loglike() == approx(-3527.68115234375)
    assert m.n_cases == 7564
    m.datatree = tree
    assert m.loglike() == approx(-8047.006193851376)
    assert m.n_cases == 20739