o 0Gf@s:dZddlZddlmZddlmZmZGdddZdS)zL Created on Sun May 10 08:23:48 2015 Author: Josef Perktold License: BSD-3 N) NonePenalty)approx_fprime_cs approx_fprimecseZdZdZfddZdddZdfdd Zdfd d Zdd d Zdfdd Z dfdd Z dddZ dfdd Z dfdd Z ZS)PenalizedMixinaMixin class for Maximum Penalized Likelihood Parameters ---------- args and kwds for the model super class penal : None or instance of Penalized function class If penal is None, then NonePenalty is used. pen_weight : float or None factor for weighting the penalization term. If None, then pen_weight is set to nobs. TODO: missing **kwds or explicit keywords TODO: do we adjust the inherited docstrings? We would need templating to add the penalization parameters cs|dd|_|dd|_tj|i||jdur"t|j|_|jdur.t|_d|_|j ddgt |dg|_ |j ddgdS)Npenal pen_weightr_null_drop_keys) poprrsuper__init__lenZendogrZ _init_keysextendgetattrr )selfargskwds __class__;lib/python3.10/site-packages/statsmodels/base/_penalized.pyr s   zPenalizedMixin.__init__NcKs2|durt|dr||}||}|Sd}|S)NZ scaletyper)hasattrZpredictZestimate_scale)rparamsscalerZmurrr _handle_scale7s   zPenalizedMixin._handle_scalec \|dur|j}tj|fi|}|dkr,|j|fi|}|d|||j|8}|S)z3 Log-likelihood of model at params Nrr)rr loglikerrfunc)rrrrllfrrrrrCzPenalizedMixin.loglikec n|dur|j}tj|fi|}t|jd}|dkr5|j|fi|}|d||||j|8}|S)z@ Log-likelihood of model observations at params Nrr)rr loglikeobsfloatshaperrr)rrrrrZnobs_llfrrrrr!Qs zPenalizedMixin.loglikeobsfdc sNdurjfdd}|dkrt||S|dkr#t||ddStd) z4score based on finite difference derivative Ncj|fdiSNrrprrrrrgz.PenalizedMixin.score_numdiff..csr$T)Zcenteredz-method not recognized, should be "fd" or "cs")rrr ValueError)rrrmethodrrrr*r score_numdiffas zPenalizedMixin.score_numdiffc r)z- Gradient of model at params Nrr)rr scorerrderiv)rrrrscrrrrr1przPenalizedMixin.scorec r )z: Gradient of model observations at params Nrr)rr score_obsr"r#rrr2)rrrrr3Znobs_scrrrrr4~s zPenalizedMixin.score_obsc s4durjfdd}ddlm}|||S)z6hessian based on finite difference derivative Ncr%r&r'r(r*rrr+r,z0PenalizedMixin.hessian_numdiff..r) approx_hess)rstatsmodels.tools.numdiffr5)rrrrrr5rr*rhessian_numdiffs   zPenalizedMixin.hessian_numdiffc s|dur|j}tj|fi|}|dkrB|j|fi|}|j|}|jdkr8|d|t||8}|S|d|||8}|S)z, Hessian of model at params Nrr) rr hessianrrZderiv2ndimnpZdiag)rrrrZhessrhrrrr8s  zPenalizedMixin.hessianc  sddlm}ddlm}t|||fr|ddi|dur d}|dur&d}tjd d|i|}|dur7|S|d ur=d }t t |j |kd}t t |j |kd}| rf|j |fi|} | S|S) aminimize negative penalized log-likelihood Parameters ---------- method : None or str Method specifies the scipy optimizer as in nonlinear MLE models. trim : {bool, float} Default is False or None, which uses no trimming. If trim is True or a float, then small parameters are set to zero. If True, then a default threshold is used. If trim is a float, then it will be used as threshold. The default threshold is currently 1e-4, but it will change in future and become penalty function dependent. kwds : extra keyword arguments This keyword arguments are treated in the same way as in the fit method of the underlying model class. Specifically, additional optimizer keywords and cov_type related keywords can be added. r)GLMGam)GLMZmax_start_irlsNZbfgsFr/Tg-C6?r)Z*statsmodels.gam.generalized_additive_modelr<Z+statsmodels.genmod.generalized_linear_modelr= isinstanceupdater fitr:ZnonzeroabsranyZ _fit_zeros) rr/Ztrimrr<r=resZ drop_indexZ keep_indexZres_auxrrrr@s&  zPenalizedMixin.fit)N)Nr$)NN)__name__ __module__ __qualname____doc__r rrr!r0r1r4r7r8r@ __classcell__rrrrr s      r) rGZnumpyr:Z _penaltiesrr6rrrrrrrs