import numpy as np

from statsmodels.tools.testing import ParamsTableTestBunch

est = dict(
           k_eq_model=0,
           phi=1,
           vf=1,
           df=28,
           df_m=3,
           power=0,
           canonical=1,
           rank=4,
           aic=1.055602138883215,
           rc=0,
           p=.0388431588742135,
           chi2=8.376256383189103,
           ll=-12.88963422213144,
           k_autoCns=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=3,
           N=32,
           nbml=0,
           bic=-71.26133683412948,
           dispers_ps=.9734684585933382,
           deviance_ps=27.25711684061347,
           dispers_p=.9734684585933382,
           deviance_p=27.25711684061347,
           dispers_s=.920688158723674,
           deviance_s=25.77926844426287,
           dispers=.920688158723674,
           deviance=25.77926844426287,
           cmdline="glm grade  gpa tuce psi, family(binomial)",
           cmd="glm",
           predict="glim_p",
           marginsnotok="stdp Anscombe Cooksd Deviance Hat Likelihood Pearson Response Score Working ADJusted STAndardized STUdentized MODified",  # noqa:E501
           marginsok="default",
           hac_lag="30",
           vcetype="OIM",
           vce="oim",
           linkt="Logit",
           linkf="ln(u/(1-u))",
           varfunct="Bernoulli",
           varfuncf="u*(1-u)",
           opt1="ML",
           oim="oim",
           a="1",
           m="1",
           varfunc="glim_v2",
           link="glim_l02",
           chi2type="Wald",
           opt="moptimize",
           title="Generalized linear models",
           user="glim_lf",
           crittype="log likelihood",
           ml_method="e2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.8261124216999,  1.2629410221647,  2.2377231969675,  .02523911156938,
    .35079350365878,   5.301431339741, np.nan,  1.9599639845401,
    0,  .09515765001172,    .141554201358,  .67223472774972,
    .50143427587633, -.18228348651028,  .37259878653373, np.nan,
    1.9599639845401,                0,  2.3786875040587,  1.0645642078703,
    2.2344237073472,  .02545520725424,  .29217999740245,  4.4651950107149,
    np.nan,  1.9599639845401,                0, -13.021345912635,
    4.931323890811, -2.6405375515688,  .00827746189686, -22.686563134726,
    -3.3561286905433, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    1.5950200254665, -.03692058012179,  .42761557297075, -4.5734780841711,
    -.03692058012179,   .0200375919221,  .01491263753083, -.34625566662867,
    .42761557297075,  .01491263753083,  1.1332969526786, -2.3591604492672,
    -4.5734780841711, -.34625566662867, -2.3591604492672,  24.317955316083
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.889634222131,                4,
    33.779268444263,  39.642212055462])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()


results_noconstraint = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                **est
                )

est = dict(
           k_eq_model=0,
           phi=1,
           vf=1,
           df=28,
           df_m=3,
           power=0,
           canonical=1,
           rank=4,
           aic=1.055602138883215,
           rc=0,
           p=.0248623136764981,
           chi2=9.360530997638559,
           ll=-12.88963422213144,
           k_autoCns=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=3,
           N=32,
           nbml=0,
           bic=-71.26133683412948,
           dispers_ps=.9734684585933382,
           deviance_ps=27.25711684061347,
           dispers_p=.9734684585933382,
           deviance_p=27.25711684061347,
           dispers_s=.920688158723674,
           deviance_s=25.77926844426287,
           dispers=.920688158723674,
           deviance=25.77926844426287,
           cmdline="glm grade  gpa tuce psi, family(binomial) vce(robust)",
           cmd="glm",
           predict="glim_p",
           marginsnotok="stdp Anscombe Cooksd Deviance Hat Likelihood Pearson Response Score Working ADJusted STAndardized STUdentized MODified",  # noqa:E501
           marginsok="default",
           hac_lag="30",
           vcetype="Robust",
           vce="robust",
           linkt="Logit",
           linkf="ln(u/(1-u))",
           varfunct="Bernoulli",
           varfuncf="u*(1-u)",
           opt1="ML",
           oim="oim",
           a="1",
           m="1",
           varfunc="glim_v2",
           link="glim_l02",
           chi2type="Wald",
           opt="moptimize",
           title="Generalized linear models",
           user="glim_lf",
           crittype="log pseudolikelihood",
           ml_method="e2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.8261124216999,   1.287827879216,  2.1944799202672,  .02820092594159,
    .30201616014984,  5.3502086832499, np.nan,  1.9599639845401,
    0,  .09515765001172,   .1198091371814,  .79424367999287,
    .42705358424294, -.13966394388263,  .32997924390608, np.nan,
    1.9599639845401,                0,  2.3786875040587,  .97985082470462,
    2.4276016757712,  .01519902587997,  .45821517741577,  4.2991598307016,
    np.nan,  1.9599639845401,                0, -13.021345912635,
    5.2807513766642, -2.4658130981467,  .01367026437574, -23.371428422207,
    -2.6712634030626, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    1.6585006464861,  .00630184631279,  .20368998146717, -5.7738061195745,
    .00630184631279,  .01435422935215,  .01997066738212, -.34768562593344,
    .20368998146717,  .01997066738212,  .96010763867432, -1.5315997267117,
    -5.7738061195745, -.34768562593344, -1.5315997267117,  27.886335102141
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.889634222131,                4,
    33.779268444263,  39.642212055462])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()


results_noconstraint_robust = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                **est
                )

est = dict(
           k_eq_model=0,
           phi=1,
           vf=1,
           df=29,
           df_m=2,
           power=0,
           canonical=1,
           rank=3,
           aic=.993115540206396,
           rc=0,
           p=.0600760311411508,
           chi2=5.624288666552698,
           ll=-12.88984864330234,
           k_autoCns=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=3,
           N=32,
           nbml=0,
           bic=-74.7266438945874,
           dispers_ps=.9340711710496038,
           deviance_ps=27.08806396043851,
           dispers_p=.9340711710496038,
           deviance_p=27.08806396043851,
           dispers_s=.8889550788484368,
           deviance_s=25.77969728660467,
           dispers=.8889550788484368,
           deviance=25.77969728660467,
           cmdline="glm grade  gpa tuce psi, family(binomial) constraints(1)",
           cmd="glm",
           predict="glim_p",
           marginsnotok="stdp Anscombe Cooksd Deviance Hat Likelihood Pearson Response Score Working ADJusted STAndardized STUdentized MODified",  # noqa:E501
           marginsok="default",
           hac_lag="30",
           vcetype="OIM",
           vce="oim",
           linkt="Logit",
           linkf="ln(u/(1-u))",
           varfunct="Bernoulli",
           varfuncf="u*(1-u)",
           opt1="ML",
           oim="oim",
           a="1",
           m="1",
           varfunc="glim_v2",
           link="glim_l02",
           chi2type="Wald",
           opt="moptimize",
           title="Generalized linear models",
           user="glim_lf",
           crittype="log likelihood",
           ml_method="e2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.8, np.nan, np.nan, np.nan,
    np.nan, np.nan, np.nan,  1.9599639845401,
    0,  .09576464077943,  .13824841412912,  .69269974185736,
    .48849800113543, -.17519727183342,  .36672655339228, np.nan,
    1.9599639845401,                0,  2.3717067235827,  1.0071435928909,
    2.3548843882081,  .01852846934254,  .39774155425619,  4.3456718929091,
    np.nan,  1.9599639845401,                0, -12.946549758905,
    3.3404275889275, -3.8757163309928,  .00010631147941, -19.493667526167,
    -6.3994319916434, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    0,                0,                0,                0,
    0,  .01911262400922,  .02461998233256, -.45036648979107,
    0,  .02461998233256,  1.0143382167012,  -1.126241119498,
    0, -.45036648979107,  -1.126241119498,  11.158456476868
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.889848643302,                3,
    31.779697286605,  36.176904995004])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()


results_constraint1 = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                **est
                )

est = dict(
           k_eq_model=0,
           phi=1,
           vf=1,
           df=29,
           df_m=2,
           power=0,
           canonical=1,
           rank=3,
           aic=.9965088127779717,
           rc=0,
           p=.0151376593316312,
           chi2=8.381139289068923,
           ll=-12.94414100444755,
           k_autoCns=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=3,
           N=32,
           nbml=0,
           bic=-74.61805917229698,
           dispers_ps=.9101961406899989,
           deviance_ps=26.39568808000997,
           dispers_p=.9101961406899989,
           deviance_p=26.39568808000997,
           dispers_s=.892699379617072,
           deviance_s=25.88828200889509,
           dispers=.892699379617072,
           deviance=25.88828200889509,
           cmdline="glm grade  gpa tuce psi, family(binomial) constraints(2)",
           cmd="glm",
           predict="glim_p",
           marginsnotok="stdp Anscombe Cooksd Deviance Hat Likelihood Pearson Response Score Working ADJusted STAndardized STUdentized MODified",  # noqa:E501
           marginsok="default",
           hac_lag="30",
           vcetype="OIM",
           vce="oim",
           linkt="Logit",
           linkf="ln(u/(1-u))",
           varfunct="Bernoulli",
           varfuncf="u*(1-u)",
           opt1="ML",
           oim="oim",
           a="1",
           m="1",
           varfunc="glim_v2",
           link="glim_l02",
           chi2type="Wald",
           opt="moptimize",
           title="Generalized linear models",
           user="glim_lf",
           crittype="log likelihood",
           ml_method="e2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.5537914524884,  .92662050289421,  2.7560273537138,  .00585081038138,
    .73764863947939,  4.3699342654975, np.nan,  1.9599639845401,
    0,  .10791139824293,  .13554656123081,  .79612051580696,
    .42596199070477, -.15775497999771,  .37357777648357, np.nan,
    1.9599639845401,                0,  2.5537914524884,  .92662050289421,
    2.7560273537138,  .00585081038138,  .73764863947939,  4.3699342654975,
    np.nan,  1.9599639845401,                0, -12.527922070831,
    4.6393777844052, -2.7003453163357,  .00692675392223,  -21.62093543894,
    -3.4349087027211, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    .85862555638391, -.00408642741742,  .85862555638391, -3.1725052764862,
    -.00408642741742,   .0183728702615, -.00408642741742, -.40376368789892,
    .85862555638391, -.00408642741742,  .85862555638391, -3.1725052764862,
    -3.1725052764862, -.40376368789892, -3.1725052764862,  21.523826226433
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.944141004448,                3,
    31.888282008895,  36.285489717294])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()

predict_mu = np.array([
    .02720933393726,  .05877785527304,  .17341537851768,  .02240274574181,
    .48834788561471,  .03262255746648,  .02545734725406,  .05057489993471,
    .10986224061161,  .64848146294279,  .02525325609066,  .17259131542841,
    .28314297612096,  .18171413480391,  .33018645131295,  .02988039105483,
    .05693576903037,  .03731338966779,  .61672273095571,  .68861137241716,
    .08792248035539,  .90822178043053,  .25295501355621,  .85758484919326,
    .83972248507748,  .54158048311843,  .62661357692624,  .36224489285202,
    .83387563062407,  .93837010344092,  .55200183830167,  .13940358008872
    ])

predict_mu_colnames = 'predict_mu'.split()

predict_mu_rownames = ['r'+str(n) for n in range(1, 33)]

predict_linpred_std = np.array([
    1.2186852972383,  .98250143329647,  .71300625338041,  1.7281112031272,
    .58278126610648,  1.2588643933597,   1.323097466817,  1.0187451680624,
    .92583226681839,  .97445803529749,  1.2426520057509,  .66674211884633,
    .53877733839827,  .77006015103931,  .70670367147137,  1.2036701125873,
    1.1407798755705,  1.1376397495763,  .57331962577752,  .65764380198652,
    .85122884445037,  1.1282943138296,  1.2981327331615,  .91561885084703,
    .8524827403359,  .75030433039358,  1.0902299962647,  .53350768600347,
    .96511132361274,  1.2127047415358,  .61923877005984,  .80300912367498
    ])

predict_linpred_std_colnames = 'predict_linpred_std'.split()

predict_linpred_std_rownames = ['r'+str(n) for n in range(1, 33)]

predict_hat = np.array([
    .03931157544567,  .05340381182541,  .07287215399916,  .06540404284993,
    .0848623883214,   .0500117280211,  .04343078449564,  .04983412818394,
    .08382437063813,  .21645722203914,  .03801090644315,  .06348261316195,
    .05891921860299,  .08817451110282,  .11045563375857,  .04199779738721,
    .06987634275981,  .04648995770552,   .0776956378885,  .09273814423054,
    .05810645039404,  .10611489289649,  .31844046474321,  .10239122636412,
    .09780916971071,  .13976583081559,  .27809589396914,  .06575633167064,
    .12902962938834,  .08505028097419,  .09482722348113,  .07735963673184
    ])

predict_hat_colnames = 'predict_hat'.split()

predict_hat_rownames = ['r'+str(n) for n in range(1, 33)]


results_constraint2 = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                predict_mu=predict_mu,
                predict_mu_colnames=predict_mu_colnames,
                predict_mu_rownames=predict_mu_rownames,
                predict_linpred_std=predict_linpred_std,
                predict_linpred_std_colnames=predict_linpred_std_colnames,
                predict_linpred_std_rownames=predict_linpred_std_rownames,
                predict_hat=predict_hat,
                predict_hat_colnames=predict_hat_colnames,
                predict_hat_rownames=predict_hat_rownames,
                **est
                )

est = dict(
           k_eq_model=0,
           phi=1,
           vf=1,
           df=29,
           df_m=2,
           power=0,
           canonical=1,
           rank=3,
           aic=.9965088127779717,
           rc=0,
           p=.0085760854232441,
           chi2=9.517555427941099,
           ll=-12.94414100444755,
           k_autoCns=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=3,
           N=32,
           nbml=0,
           bic=-74.61805917229698,
           dispers_ps=.9101961406899989,
           deviance_ps=26.39568808000997,
           dispers_p=.9101961406899989,
           deviance_p=26.39568808000997,
           dispers_s=.892699379617072,
           deviance_s=25.88828200889509,
           dispers=.892699379617072,
           deviance=25.88828200889509,
           cmdline="glm grade  gpa tuce psi, family(binomial) constraints(2) vce(robust)",  # noqa:E501
           cmd="glm",
           predict="glim_p",
           marginsnotok="stdp Anscombe Cooksd Deviance Hat Likelihood Pearson Response Score Working ADJusted STAndardized STUdentized MODified",  # noqa:E501
           marginsok="default",
           hac_lag="30",
           vcetype="Robust",
           vce="robust",
           linkt="Logit",
           linkf="ln(u/(1-u))",
           varfunct="Bernoulli",
           varfuncf="u*(1-u)",
           opt1="ML",
           oim="oim",
           a="1",
           m="1",
           varfunc="glim_v2",
           link="glim_l02",
           chi2type="Wald",
           opt="moptimize",
           title="Generalized linear models",
           user="glim_lf",
           crittype="log pseudolikelihood",
           ml_method="e2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.5537914524884,  .83609404798719,  3.0544308485827,  .00225487991353,
    .91507723074524,  4.1925056742316, np.nan,  1.9599639845401,
    0,  .10791139824293,  .12275592600281,   .8790728216287,
    .37936179287834, -.13268579561143,   .3485085920973, np.nan,
    1.9599639845401,                0,  2.5537914524884,  .83609404798719,
    3.0544308485827,  .00225487991353,  .91507723074524,  4.1925056742316,
    np.nan,  1.9599639845401,                0, -12.527922070831,
    4.510414281113, -2.7775546302454,  .00547696322683, -21.368171617167,
    -3.6876725244938, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    .6990532570796,  .01512804251258,   .6990532570796, -2.9662622048441,
    .01512804251258,  .01506901736881,  .01512804251258,  -.3968065659911,
    .6990532570796,  .01512804251258,   .6990532570796, -2.9662622048441,
    -2.9662622048441,  -.3968065659911, -2.9662622048441,  20.343836987269
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.944141004448,                3,
    31.888282008895,  36.285489717294])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()


results_constraint2_robust = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                **est
                )

est = dict(
           N_cds=0,
           N_cdf=0,
           p=.0151376589433054,
           chi2=8.381139340374848,
           df_m=2,
           k_eq_model=1,
           ll=-12.94414100444751,
           k_autoCns=0,
           rc=0,
           converged=1,
           k_dv=1,
           k_eq=1,
           k=4,
           ic=5,
           N=32,
           rank=3,
           cmdline="logit grade  gpa tuce psi, constraints(2)",
           cmd="logit",
           estat_cmd="logit_estat",
           predict="logit_p",
           marginsnotok="stdp DBeta DEviance DX2 DDeviance Hat Number Residuals RStandard SCore",  # noqa:E501
           title="Logistic regression",
           chi2type="Wald",
           opt="moptimize",
           vce="oim",
           user="mopt__logit_d2()",
           crittype="log likelihood",
           ml_method="d2",
           singularHmethod="m-marquardt",
           technique="nr",
           which="max",
           depvar="grade",
           properties="b V",
          )

params_table = np.array([
    2.5537916456996,  .92662056628814,  2.7560273736742,  .00585081002433,
    .73764870844071,  4.3699345829585, np.nan,  1.9599639845401,
    0,  .10791141442743,  .13554656655573,  .79612060393329,
    .42596193948753, -.15775497424986,  .37357780310472, np.nan,
    1.9599639845401,                0,  2.5537916456996,  .92662056628814,
    2.7560273736742,  .00585081002433,  .73764870844071,  4.3699345829585,
    np.nan,  1.9599639845401,                0, -12.527923225554,
    4.6393781670436, -2.7003453425175,  .00692675337706,  -21.62093734362,
    -3.4349091074867, np.nan,  1.9599639845401,                0
    ]).reshape(4, 9)

params_table_colnames = 'b se z pvalue ll ul df crit eform'.split()

params_table_rownames = 'gpa tuce psi _cons'.split()

cov = np.array([
    .85862567386816, -.00408642236043,  .85862567386816,  -3.172505858545,
    -.00408642236043,  .01837287170505, -.00408642236043, -.40376374127778,
    .85862567386816, -.00408642236043,  .85862567386816,  -3.172505858545,
    -3.172505858545, -.40376374127778,  -3.172505858545,  21.523829776841
    ]).reshape(4, 4)

cov_colnames = 'gpa tuce psi _cons'.split()

cov_rownames = 'gpa tuce psi _cons'.split()

infocrit = np.array([
    32, np.nan, -12.944141004448,                3,
    31.888282008895,  36.285489717294])

infocrit_colnames = 'N ll0 ll df AIC BIC'.split()

infocrit_rownames = '.'.split()


results_logit_constraint2 = ParamsTableTestBunch(
                params_table=params_table,
                params_table_colnames=params_table_colnames,
                params_table_rownames=params_table_rownames,
                cov=cov,
                cov_colnames=cov_colnames,
                cov_rownames=cov_rownames,
                infocrit=infocrit,
                infocrit_colnames=infocrit_colnames,
                infocrit_rownames=infocrit_rownames,
                **est
                )