o 0GfI@sdZddlZddlmmZddlmmZ ddl mm Z ddl mZmZddlZddlZddlZGdddejZGdddeZGdd d eZGd d d ejZGd d d eZGddde jZe eedS)zA Conditional logistic, Poisson, and multinomial logit regression N)MultinomialResultsMultinomialResultsWrappercsneZdZdfdd ZddZ     dfd d Z    dddZe  dfdd ZZ S)_ConditionalModelnonec sd|vrtd|d}|j|jkrd}t||jd|jkr&d}t|tj||fd|i||jjdur?d}t||j}|jd|_i}t |D]\}} | |vrZg|| <||  |qNt |t |}}| d } g|_g|_g|_| durt | } g|_g|_g|_d|_ddg} |D][\} } || j} t | dkr| dd7<| dt| 7<q|jt| 7_|j | | dur|j | | |j t| |j || ddf|j t | q| ddkrd t| }t|| dur&g|_t |jD]\}}|j t |j||qt|j|_g|_ g|_!t"|jD]} |j t |j| |j| |j! t |j| q7dS) Ngroups'groups' is a required argumentz4'endog' and 'groups' should have the same dimensionsrzBThe leading dimension of 'exog' should equal the length of 'endog'missingzDConditional models should not have an intercept in the design matrixoffsetzIDropped %d groups and %d observations for having no within-group variance)# ValueErrorsizeshapesuper__init__dataZ const_idxexogk_params enumerateappendnpZasarrayget _endog_grp _exog_grp _groupsize _offset_grpZ_offset_sumynobsitemsZflatZstdlensumtuplewarningswarn_endofsdot _n_groups_xy_n1range)selfendogrrkwargsrmsgZrow_ixigr ZdropsZixykofs __class__Glib/python3.10/site-packages/statsmodels/discrete/conditional_models.pyrs           z_ConditionalModel.__init__cCs&ddlm}|||j}t|}|S)Nr) approx_fprime)Zstatsmodels.tools.numdiffr6scorerZ atleast_2d)r)paramsr6Zhessr4r4r5hessianbs   z_ConditionalModel.hessianNBFGSdTFr4c  sztj|||||| d} t|| j| d} || _|j| _|j| _dt |j dt |j dt |j g| _t| } | S)N start_paramsmethodmaxiter full_outputdisp skip_hessianr z%dz%.1f)rfitConditionalResultsr8Z cov_paramsr>rr%n_groupsminrmaxrZmean _group_statsConditionalResultsWrapper) r)r=r>r?r@rAfargscallbackretallrBr+rsltZcrsltr2r4r5rCis$   z_ConditionalModel.fit elastic_netcKsNddlm}|dkrtdddddd}||||f||||d |S) a Return a regularized fit to a linear regression model. Parameters ---------- method : {'elastic_net'} Only the `elastic_net` approach is currently implemented. alpha : scalar or array_like The penalty weight. If a scalar, the same penalty weight applies to all variables in the model. If a vector, it must have the same length as `params`, and contains a penalty weight for each coefficient. start_params : array_like Starting values for `params`. refit : bool If True, the model is refit using only the variables that have non-zero coefficients in the regularized fit. The refitted model is not regularized. **kwargs Additional keyword argument that are used when fitting the model. Returns ------- Results A results instance. r)fit_elasticnetrNz.method for fit_regularized must be elastic_net2r g|=)r?ZL1_wtZ cnvrg_tolZzero_tol)r>alphar=refit)Zstatsmodels.base.elastic_netrPr update)r)r>rRr=rSr+rPdefaultsr4r4r5fit_regularizeds ! z!_ConditionalModel.fit_regularizedc szz |d}|d=Wn tytdwt|tr||}d|ddvr+tdtj|g|R||d|}|S)Nrrz0+ z2Conditional models should not include an intercept)rr) KeyErrorr isinstancestrreplacer!r"r from_formula) clsZformularZsubsetZ drop_colsargsr+rmodelr2r4r5r]s(     z_ConditionalModel.from_formular Nr:r;TFr4NFF)rNrONF)NN) __name__ __module__ __qualname__rr9rCrV classmethodr] __classcell__r4r4r2r5rs,P! 1rcsVeZdZdZdfdd ZddZddZdd d Zdd d ZddZ ddZ Z S)ConditionalLogita Fit a conditional logistic regression model to grouped data. Every group is implicitly given an intercept, but the model is fit using a conditional likelihood in which the intercepts are not present. Thus, intercept estimates are not given, but the other parameter estimates can be interpreted as being adjusted for any group-level confounders. Parameters ---------- endog : array_like The response variable, must contain only 0 and 1. exog : array_like The array of covariates. Do not include an intercept in this array. groups : array_like Codes defining the groups. This is a required keyword parameter. rc sTtj||fd|i|tt|jtjdkr!d}t||jj d|_ dS)Nr)rr zendog must be coded as 0, 1r ) rrranyuniquer*Zr_r rr K)r)r*rrr+r,r2r4r5rszConditionalLogit.__init__cCs,d}tt|jD] }||||7}q |SNr)r(rr loglike_grp)r)r8llr.r4r4r5loglikeszConditionalLogit.loglikecCs(d}t|jD] }||||7}q|Srl)r(r% score_grp)r)r8r7r.r4r4r5r7szConditionalLogit.scoreNcsR|durd}tt|j|||ifdd|j||j|S)Nrcsv||krdS|dkr dSz||fWStyYnw|d||d|d|d}|||f<|S)Nrr )rY)tr0vexbfmemor4r5rus , z"ConditionalLogit._denom..f)rexpr$rrr'r)grpr8r1r4rsr5_denoms zConditionalLogit._denomcsZ|durd}j|tt||ifddj|j|S)Nrc s||kr dtjfS|dkrdSz||fWSty"Ynw|d}|d|\}}|d|d\}}|||dddf}|||||||}} || f||f<|| fS)Nr)r rr )rzerosrrY) rqr0habcedurrexrtrvsr)r4r5r.s   z'ConditionalLogit._denom_grad..s)rrrwr$rr'rxr4rr5 _denom_grad"s zConditionalLogit._denom_gradcCs\d}t|dr |j|}t|j||}|dur ||j|7}|t||||8}|S)Nr )hasattrrrr$r&r#logrz)r)ryr8r1Zllgr4r4r5rmFs  zConditionalLogit.loglike_grpcCs<d}t|dr |j|}||||\}}|j|||S)Nrr )rrrr&)r)ryr8r1rr|r4r4r5rpUs   zConditionalLogit.score_grpra)N) rcrdre__doc__rror7rzrrmrprgr4r4r2r5rhs   $rhc@s eZdZdZddZddZdS)ConditionalPoissonaU Fit a conditional Poisson regression model to grouped data. Every group is implicitly given an intercept, but the model is fit using a conditional likelihood in which the intercepts are not present. Thus, intercept estimates are not given, but the other parameter estimates can be interpreted as being adjusted for any group-level confounders. Parameters ---------- endog : array_like The response variable exog : array_like The covariates groups : array_like Codes defining the groups. This is a required keyword parameter. c Csd}t|dr |j}d}tt|jD]7}t|j||}|dur(|||7}t|}|j|}|t||7}| }||j |t |8}q|SNr rO) rrr(rrrr$rrwrrr) r)r8r1rnr-xbrtr/rr4r4r5rors    zConditionalPoisson.loglikec Csd}t|dr |j}d}tt|jD]<}|j|}t||}|dur*|||7}t|}| }|j|} |t| |7}||j |t|||8}q|Sr) rrr(rrrrr$rwrr) r)r8r1r7r-xrrtrr/r4r4r5r7s       zConditionalPoisson.scoreN)rcrdrerror7r4r4r4r5r_s rcs&eZdZfddZdddZZS)rDcstj||||ddS)N)normalized_cov_paramsscale)rr)r)r`r8rrr2r4r5rs  zConditionalResults.__init__N皙?c Csdddd|jgfddg}dd|jgfd |jd gfd |jd gfd |jdgfg}|dur/d}d dlm}|}|j||||||d|j|||||jd|S)a< Summarize the fitted model. Parameters ---------- yname : str, optional Default is `y` xname : list[str], optional Names for the exogenous variables, default is "var_xx". Must match the number of parameters in the model title : str, optional Title for the top table. If not None, then this replaces the default title alpha : float Significance level for the confidence intervals Returns ------- smry : Summary instance This holds the summary tables and text, which can be printed or converted to various output formats. See Also -------- statsmodels.iolib.summary.Summary : class to hold summary results )zDep. Variable:N)zModel:N)zLog-Likelihood:NzMethod:)zDate:N)zTime:N)zNo. Observations:Nz No. groups:zMin group size:rzMax group size:r zMean group size:Nz*Conditional Logit Model Regression Results)Summary)ZgleftZgrightynamexnametitle)rrrRuse_t)r>rErHZstatsmodels.iolib.summaryrZadd_table_2colsZadd_table_paramsr) r)rrrrRZtop_leftZ top_rightrZsmryr4r4r5summarys:    zConditionalResults.summary)NNNr)rcrdrerrrgr4r4r2r5rDs rDcsNeZdZdZdfdd Z      dd d Zd dZddZZS)ConditionalMNLogita Fit a conditional multinomial logit model to grouped data. Parameters ---------- endog : array_like The dependent variable, must be integer-valued, coded 0, 1, ..., c-1, where c is the number of response categories. exog : array_like The independent variables. groups : array_like Codes defining the groups. This is a required keyword parameter. Notes ----- Equivalent to femlogit in Stata. References ---------- Gary Chamberlain (1980). Analysis of covariance with qualitative data. The Review of Economic Studies. Vol. 47, No. 1, pp. 225-238. rc stj||fd|i||jt|_|jd|_|jd|jjd|_ |j |j |_ ddt |jD|_ |j|_|jjd|_|jdkrRd}t|ttt|jD] \}}||q\t|_|jfdd|jD|_dS) Nrr cSsi|]}|t|qSr4)r[).0jr4r4r5 sz/ConditionalMNLogit.__init__..rz%endog may not contain negative valuescsg|]}|qSr4r4)rr0Zgrxr4r5 sz/ConditionalMNLogit.__init__..)rrr*ZastypeintrGk_catrr Zdf_modelrZdf_residr(Z _ynames_mapJrkrFr collections defaultdictlistrrrkeysZ _group_labelssort_grp_ix)r)r*rrr+r,r0rrr2rr5rs.  zConditionalMNLogit.__init__Nr:r;TFr4c  Ks|dur|jjd} |jd} tjj| | d}tjj||||||| d} | j |jjddf| _ t || } | j tj dt| S)Nr )r r<)Zllnull)rr rrZrandomZnormalbaseLikelihoodModelrCr8reshaperZset_null_optionsnanr)r)r=r>r?r@rArJrKrLrBr+qrrMr4r4r5rCs"   zConditionalMNLogit.fitc Cs|jjd}|jd}|||f}tjt|df|fdd}t|j|}d}|jD]?}||ddf}tj |jdt d} |j |} d} t | D]} | t|| | f7} qL||| | ft| 7}q,|S)Nr ZaxisrOrZdtype)rr rrr concatenater{r$rarangerr* itertools permutationsrwrr) r)r8rrpmatlprrniirjjr/denompr4r4r5ro?s     zConditionalMNLogit.loglikec Cs~|jjd}|jd}|||f}tjt|df|fdd}t|j|}t||f}|jD]}||ddf}tj |jdt d} |j |} d} t||f} t | D]7} t|| | f}| |7} t| D]!\}}|dkr| dd|df||j||ddf7<qmqXt| D]\}}|dkr|dd|df|j||ddf7<q|| | 8}q1|S)Nr rrrrO)rr rrrrr{r$rrrr*rrrwrrZflatten)r)r8rrrrZgradrrrr/rZdenomgrrrr-rr4r4r5r7Us4    2.zConditionalMNLogit.scorerarb) rcrdrerrrCror7rgr4r4r2r5rs %rc@s eZdZdS)rIN)rcrdrer4r4r4r5rIvsrI)rZnumpyrZstatsmodels.base.modelrr`Z#statsmodels.regression.linear_modelZ regressionZ linear_modelZlmZstatsmodels.base.wrapperwrapperwrapZ#statsmodels.discrete.discrete_modelrrrr!rrrrhrZLikelihoodModelResultsrDrZRegressionResultsWrapperrIZpopulate_wrapperr4r4r4r5s&F ?I