U mdٚ@sdZddlZddlZddlmZmZddlZddl Z ddl m Z ddl mZddlmZddlmZddlmZdd lmZmZeZeeje_eeje_eejddd fejddd f<eejd d e_Gd ddZGdddeZ GdddeZ!GdddeZ"GdddeZ#GdddeZ$GdddeZ%GdddeZ&GdddeZ'Gdd d eZ(Gd!d"d"eZ)dId#d$Z*d%d&Z+d'd(Z,Gd)d*d*eZ-Gd+d,d,eZ.Gd-d.d.eZ/Gd/d0d0eZ0Gd1d2d2eZ1Gd3d4d4eZ2d5d6Z3dddd7d7d7d d d dddd7d7d7d d gZ4e j5j6d8e4ee4e7e4gd9d:d;gd<d=d>Z8d?d@Z9GdAdBdBeZ:GdCdDdDeZ;dEdFZ>atolrtol corr_factr normalized_cov_paramsg:0yE>Hz>)res1res2rparamsgetattrhasattrr isinstanceTestRepeatedvsAggregatedTestRepeatedvsAverageTestTweedieRepeatedvsAggregatedTestTweedieRepeatedvsAverageTestBinomial0RepeatedvsAverage!TestBinomial0RepeatedvsDuplicatedbseTestBinomialVsVarWeightsTestGlmGaussianWLSTestGlmGaussianAwNrTestGlmGammaAwNrllflldeviance)selfrrrr-b/home/sam/Atlas/atlas_env/lib/python3.8/site-packages/statsmodels/genmod/tests/test_glm_weights.py test_basic8s2    zCheckWeight.test_basiccCst|ttttttfrdS|j}|j}t |ds4dSt t |j |j j}t|j|ddddt|j|ddddt|j|ddddt|j|dddd|d dkrdS|j}|d t|j}t||ddddS) Nresidsresid_responserrr resid_pearsonresid_deviance resid_workingresid_anscombe)rrrr r!r"r#rrrdictzipZresids_colnamesr0Trr1r2r3r4getr5npsqrt _var_weights)r,rrZ resid_allZresid_aZresid_a1r-r-r.test_residualsVsB     zCheckWeight.test_residualscCs|j}t|jjtjjr"d}d}nd}d}t|ttt t fr@dS|j }|jjj |||d}t |j |j ddd|jj|j d d }tttj| }t |j|ddddS) NnewtonZeimbfgsZoim) start_paramsmethod optim_hessianMbP?gMb`?rFZobserved)rrmodelfamilysmfamiliesTweedieTestGlmPoissonFwHCTestGlmPoissonAwHCTestGlmPoissonFwClur"rfitrhessianr:r;diaglinalginvr$)r,rrArBr@rHZres2_bser-r-r.test_compare_optimizersus& z#CheckWeight.test_compare_optimizerscCs(t|jdr$t|jj|jjddddS)NZchi2rr)rrrrZ pearson_chi2Z deviance_pr,r-r-r.test_pearson_chi2s  zCheckWeight.test_pearson_chi2cCs$|j}t|jj|jjdddS)N绽|=)r)rZget_predictionrZlinpredZse_mean)r,predr-r-r.test_getpredictions zCheckWeight.test_getpredictionN)__name__ __module__ __qualname__r/r=rSrUrXr-r-r-r.r7s rc@seZdZeddZdS)TestGlmPoissonPlaincCs*ttjtjtjd|_t j |_ dS)NrF) r cpunish_dataendogexogrGrHPoissonrMr res_stataZresults_poisson_none_nonrobustr)clsr-r-r. setup_classs  zTestGlmPoissonPlain.setup_classNrYrZr[ classmethodrdr-r-r-r.r\sr\c@seZdZeddZdS)TestGlmPoissonFwNrcCs\dddddddddddddddddg}t|}ttjtjtj|d |_ t j |_ dS)Nr r rF freq_weights)r:arrayrr^r_r`rGrHrarMrrbZ!results_poisson_fweight_nonrobustr)rcfweightsr-r-r.rds&   zTestGlmPoissonFwNr.setup_classNrer-r-r-r.rgsrgc@seZdZeddZdS)TestGlmPoissonAwNrcCsdddddddddddddddddg}t|}|}ttj}|||}ttjtjtj |d |_ ddl m }|tj|_|jj |j_|jjddddft|ddtjf9<dS)Nr rhr rF var_weightsr)copy)r:rksumlenr^r_rr`rGrHrarMrrprbZ!results_poisson_aweight_nonrobustrr0r;Znewaxis)rcrlwsumnobsaweightsrpr-r-r.rds&       zTestGlmPoissonAwNr.setup_classNrer-r-r-r.rmsrmcsTeZdZeddZejjdddfddZejjdddfdd Z Z S) TestGlmPoissonPwNrcCs~dddddddddddddddddg}t|}|}ttj}|||}ttjtjtj |dj dd|_ t j|_dS)Nr rhr riZHC1cov_type)r:rkrrrsr^r_rr`rGrHrarMrrbZ!results_poisson_pweight_nonrobustr)rcrlrtrurvr-r-r.rds&    zTestGlmPoissonPwNr.setup_classz Known to failT)reasonstrictcstt|dSN)superrwr/rT __class__r-r.r/szTestGlmPoissonPwNr.test_basiccstt|dSr|)r}rwrSrTr~r-r.rSsz*TestGlmPoissonPwNr.test_compare_optimizers) rYrZr[rfrdpytestmarkZxfailr/rS __classcell__r-r-r~r.rws  rwc@seZdZeddZdS)rJcCsdddddddddddddddddg}t|}|}ttj}|||}t|d||_ttjtj t j |d}|j dd|_tj|_dS)Nr rhr ?riHC0rx)r:rkrrrsr^r_r;rrr`rGrHrarMrrbZresults_poisson_fweight_hc1rrcrlrtrurvmodr-r-r.rds&    zTestGlmPoissonFwHC.setup_classNrer-r-r-r.rJsrJc@seZdZeddZdS)rKcCsdddddddddddddddddg}t|}|}ttj}|||}t|d|d|_ttjtj t j |d}|j dd|_tj|_dS) Nr rhr rgL#?rnrrx)r:rkrrrsr^r_r;rrr`rGrHrarMrrbZresults_poisson_aweight_hc1rrr-r-r.rds&    zTestGlmPoissonAwHC.setup_classNrer-r-r-r.rKsrKc@seZdZeddZdS)rLc Csdddddddddddddddddg}t|}|}ttj}|||}tddd}tt|}dt||d|_ |dd}t t 0t tjtjtj|d}|jd|d |_W5QRXtj|_dS) Nr rhr FgroupsZuse_correctionriclusterrycov_kwds)r:rkrrrsr^r_arangeuniquer;rrwarnsrrr`rGrHrarMrrbZresults_poisson_fweight_clu1r) rcrlrtrurvgidZn_groupsrrr-r-r.rd s &      zTestGlmPoissonFwClu.setup_classNrer-r-r-r.rL srLc@seZdZeddZdS)TestGlmTweedieAwNrcCsddlmm}tjj}|j}|j}||d<t dt |j }d|ddd<d|ddd<|j d|tjjdtjjd |d }|jd dd |_tj|_dS) Nrfairr rqr zfair ~ age + yrs_marriedg?) var_powerlinkdatarFro}:rr)statsmodels.formula.apiformulaapirGdatasetsr load_pandasr_r`r:repeatrsindexglmrHrIlinksLogrMrres_rZ"results_tweedie_aweights_nonrobustrrcsmfrr_rvrEr-r-r.rd$s&   zTestGlmTweedieAwNr.setup_classNrer-r-r-r.r#src@s eZdZeddZddZdS)r(cCsddlm}|}|j}|jddddf}tj|dd}tdt|}d|ddd<d|ddd<tj ||tj j tj j d |d }|jd d d |_tj|_dS)Nr ) CancerLogTr rqr rrrnrrr)Zresults.results_glmrr_r`rGrr:rrsrrHGammarrrMrrZ results_gamma_aweights_nonrobustr)rcrrr_r`rvrEr-r-r.rd=s zTestGlmGammaAwNr.setup_classcCsP|jj|jj}|jjj|jjj|jj|jj |d}t ||j j ddddS)Nrjscalerrr) rr+ _iweightsrrrFloglikerEr_mur<rrr*)r,rr*r-r-r. test_r_llfNszTestGlmGammaAwNr.test_r_llfNrYrZr[rfrdrr-r-r-r.r(<s r(c@s eZdZeddZddZdS)r'cCsddlmm}tjj}|j}|j}||d<|dd<t ddddd dddddddd ddddg}|j d |tj j tj jd |d }|jd dd|_tj|_dS)Nr EXECUTIONSINCOMEr rhr rqEXECUTIONS ~ INCOME + SOUTH - 1rrrr)rrrrGrcpunishrr_r`r:rkrrHGaussianrrrMrrZ#results_gaussian_aweights_nonrobustrrr-r-r.rdXs" $zTestGlmGaussianAwNr.setup_classc Cs|j}|j}|jj}|j|j|j}|j}|jj|j |j ||d}d|j |j d |}|j dt t |jd}|||} t| |jddddS)Nrgrhrrr)rrrErZdf_residZwnobsrjrFrr_rrrr:logrorr*) r,rrrErwtsr)Zadj_smZadj_rZllf_adjr-r-r.rls" zTestGlmGaussianAwNr.test_r_llfNrr-r-r-r.r'Ws r'c Cstjdtj}||}||jkr|dkrLdtjjt|d|k}nnt t|df}d}tjjt||fd|dddfk d|dddf<||dddf|dddf<n||j krtj |}n||j krtjd|}n||jkr|tjjt|d}n||jkr8ddlm}||d}np||jkr\dd lm} | |}nL||jkrd} d } || | } tjj | | jddtj| | }nt|S) Nihrr sizerh )nbinom?)invgaussr)r:randomseedrGrHZinverseBinomialuniformrsemptyrrrapoissonrgammarnormalNegativeBinomialZscipy.stats.distributionsrZrvsInverseGaussianrrIshape ValueError) lin_pred family_classr binom_versionfamrr_nrrZraterrr-r-r. gen_endogsF   "          rc Cstjdtj}tjj}|j|j|j|j |j |j gf|j |j |j |jgf|j|j |j |jgf|j|j |j |jgf|j|j |j |j|jgf|j|j |j|j|j gfg}d}d}tjj||fd}d|dddf<d}|D]\}}|D]} dD]} d } ||jkr| dkrqn||jkr4| |j kr4qn||jkrR| |j krRqn|| f|j |j fkrzd |d} nn|| f|j|j fkrd |dd } nB|| f|j |jfkrd |d} n|| f|j|j fkrqn|| f|j|j fkrqn|| f|j|jfkr"qn|| f|j|j fkr>qn|| f|j|jfkrZqn|| f|j|j fkrd |d} qnd|| f|j|j fkrd d|d} t| dtj} qn&|| f|j|jfkrd|dd} qn|| f|j|jfkrd|dd} d} n|| f|j|j fkrZd d|d} t| dtj} d} n|| f|j|jfkrdtjj|jdd} qn\|| f|j|jfkrd|dd} qn0|| f|j|jfkrd}ntjj|jdd} t| || | } | dkr:t| }tjjdd| | kd}||| | k<nZtj| jdd}| dddf| jdd}tjj d||kd}||||k<t!"*t!#dtj$| |||| dd}W5QRX|j%dddd}d|j&fdfD]\}}|dkr|rqt!"*t!#dtj$| |||| dd}W5QRX|j%||| d }t'|j&|j&d!d"d#t'|j(|j(d!d!d#t'|j)|j)d!d!d#|j*}|j+|j&dd$}t,t-tj./| }t'||j*d!d"d#qqqqdS)%Ni.Udr rr rF)rr r?rrqg-C6?rr>rC皙?TrhrZaxisignorer)rorFZIRLSrVr)rAr tol_criterion)r N)max_start_irlsr@rArg-C6 ?rrD)0r:rrrGrHrrLogitZProbitZCLogLogrZCauchyraIdentityZSqrtrZ InversePowerrrZInverseSquaredrrrrZclipinfrrrZ ones_likerandintZmeanZonesrwarningscatch_warnings simplefilterrrMrrr)rr$rNr;rOrPrQ)rZlnkrHrpr`Zskip_onerZ family_linksrrrArr_rtmpyZmod_irlsZ rslt_irlsrr@Z mod_gradientZ rslt_gradientZ gradient_bseZehessr-r-r.test_wtd_gradient_irlss                           rcCsptt|}tj|jdt|dfd}t|D]4\}}|dkrHq6t||kdd|dd|df<q6|S)Nrr r)r:sortrZzerosrrs enumeratewhere)xvaluesoutivr-r-r. get_dummiesDs$rc@seZdZeddZdS)rcCstjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||} t j| |||dd} | |_t|} | | d<| dddd d gdg} | dddd d gdg} t| j}| d} | d} t j| |||d| d}||_dS)Nrrqr rlowhighrrhr rrr皙皙?ffffff?rrr]r_rFexposurer:rrrrrrkrrrrGrHrarrrrrMrpd DataFramegroupbycountrtolistrrcrrr`rbetarrFrr_mod1agg agg_endogagg_wtagg_exogmod2r-r-r.rdOs4     z$TestRepeatedvsAggregated.setup_classNrer-r-r-r.rNsrc@seZdZeddZdS)rcCstjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||} t j| |||dd} | |_t|} | | d<| dddd d gdg} | dddd d gdg} t| j}| d} | d| }t j||||d| d}||_dS)Nr'rqr rrrrhr rrrrrrrrr]r_rnrrcrrr`rrrrFrr_rrrrrZ avg_endogrr-r-r.rdns4      z!TestRepeatedvsAverage.setup_classNrer-r-r-r.rmsrc@seZdZeddZdS)r cCstjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||} t j| |||ddd} | jdddd|_t|} | | d<| dddd d gdg} | dddd d gdg} t| j}| d} | d} t j| |||dd| | dd}|jdddd|_dS)Nrr rqr rrrrhr rrrrrr?rrr]g#B ;rrrrr_r)rFrror:rrrrrrkrrrrGrHrIrrrrrMrrrrrrrrrr-r-r.rds8     z+TestTweedieRepeatedvsAggregated.setup_classNrer-r-r-r.r sr c@seZdZeddZdS)r!cCstjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||} t j| |||ddd} | jddddd|_t|} | | d<| dddd d gdg} | dddd d gdg} t| j}| d} | d| }t j||||dd| d}|jdddd|_dS)Nrrrqr rrrrhr rr rrrrrr r r]rVrx2rrrZ scaletyper_rnrrr r-r-r.rds:       z(TestTweedieRepeatedvsAverage.setup_classNrer-r-r-r.r!sr!c@seZdZeddZdS)r"cCstjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||dd} t j| |||dd} | jddddd|_t|} | | d<| dddd d gdg} | dddd d gdg} t| j}| d} | d| }t j||||d| d}|jdddd|_dS)Nrrrqr rrrrhr rrrrrrrrrr]rVrrrr_rnr)r:rrrrrrkrrrrGrHrrrrrrMrrrrrrrrr r-r-r.rds:      z*TestBinomial0RepeatedvsAverage.setup_classNrer-r-r-r.r"sr"c@seZdZeddZdS)r#cCsBtjdd}d}t||f}d|dddf<tjjdd|d|dddf<ttddd d g|d }t||ddddf<td d d ddg}||jdd}t j j }t j j j }t|||dd} tjdd|} t j| |||d| d} | |_tj|| dd} t| | } t j| | ||dd}||_dS)Nrr rqr rrrrhr rrrrrrrrrrir])r:rrrrrrkrrrrGrHrrrrrrMrr)rcrrr`rrrrFrr_wtrZexog_dupZ endog_duprr-r-r.rds(    z-TestBinomial0RepeatedvsDuplicated.setup_classNrer-r-r-r.r#sr#cCsdddddddddddddddddg}t|}tddd}|dd}tt2ttjtj t j |dj d|d }|W5QRXtt2ttjtj t j |d j d|d }|W5QRXdS) Nr rhr rFrrirrrn)r:rkrrrrrr^r_r`rGrHrarMsummary)weightsrrrr-r-r.test_warnings_raiseds,&      rrh formattedlistndarraySeries)ZidscCs t|dSr|)check_weights_as_formats)rr-r-r.test_weights_different_formats!srcCsttjtjtj|d}t|j t j s0t t|j t j sBt t|jt j sTt ttjtjtj|d}t|j t j st t|j t j st t|jt j st dS)Nrirn)rr^r_r`rGrHrarMrZ _freq_weightsr:rAssertionErrorr<r)rresr-r-r.r(s    rc@seZdZeddZdS)r%cCsddlm}|}tj|jdd|_tj|jdd|_|j|jd_t|jdd|_t|j|jt j d |_ |jjdd }|jdddf|}t||jt j |d  |_dS) NrrW) requirementsFr r]r rrn)Zstatsmodels.datasets.star98rr:requirer`r_ZstdrrrGrHrrMrrrr)rcrrrZendog2r-r-r.rd9s   z$TestBinomialVsVarWeights.setup_classNrer-r-r-r.r%8sr%c@seZdZeddZdS)r&cCsddlmm}tjj}|j}|j}||d<|dd<t ddddd dddddddd ddddg}|j d |tj j tj jd |d }|jd ||d }|jddd|_||_dS)Nrrrrr rhr rrqrrr)rrrr)rrrrGrrrr_r`r:rkrrHrrrZwlsrMrr)rcrrr_rvrEZwlsmodelr-r-r.rdLs, $zTestGlmGaussianWLS.setup_classNrer-r-r-r.r&Ksr&cCsdddddddddddddddddg}tj}tj}tj}ttt||||dddttt||||dddttt||||dgdttt||||dgdttt|||||gdttt|||||gddS)Nr rhr rrirn) r^r`r_rGrHrarrr)rr`r_rFr-r-r.test_incompatible_inputes,&         r"c Csd\}}tjdtj||d}t|}|dd}|dtj|}dt|d}tjt||}|ddd7<t j }t ||||d}| } t |||||d } | } t| j|| jt| j| jt| j| jt| j| jt| j| jdS) N)rrqir rhrrrrn)r:rrZrandnrrrrrexprGrHrarrMrr1r2r3r5Zresid_anscombe_unscaled) ruZk_exogrZy_truerrZyprZ mod_poi_eZ res_poi_eZ mod_poi_wZ res_poi_wr-r-r.test_poisson_residuals|s.   r%)r)>__doc__rnumpyr:Z numpy.testingrrZpandasrrZstatsmodels.apirrGZstatsmodels.datasets.cpunishrZ+statsmodels.genmod.generalized_linear_modelrZstatsmodels.tools.sm_exceptionsrZstatsmodels.tools.toolsrresultsr rr rbr^Zasarrayr`r_rrr\rgrmrwrJrKrLrr(r'rrrrrr r!r"r#rrrZ parametrizerrrr%r&r"r%r-r-r-r.sd     &]  + )  !!&