M/Phm#dZddlZddlZddlmZddlmZddlm Z m Z m Z m Z m Z mZGddZGdd eZdS) zM Created on Wed Nov 18 15:17:58 2020 Author: Josef Perktold License: BSD-3 N)cache_readonly)test_chisquare_binning)test_poisson_dispersiontest_poisson_zeroinflation_jh test_poisson_zeroinflation_broektest_poisson_zerostest_chisquare_prob plot_probscDeZdZdZddZedZd dZ d dZdS) CountDiagnostica?Diagnostic and specification tests and plots for Count model status: experimental Parameters ---------- results : Results instance of a count model. y_max : int Largest count to include when computing predicted probabilities for counts. Default is the largest observed count. Nc"||_||_dSN)resultsy_max)selfrrs _/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/statsmodels/discrete/diagnostic.py__init__zCountDiagnostic.__init__+s  cz|jdtj|jdzi}ni}|jjdddi|S)Ny_valueswhichprob)rnparangerpredict)rkwdss rprobs_predictedzCountDiagnostic.probs_predicted/sK : ! $*q. 9 9:DDD#t|#99&9D999rci}| tj|ddz|d<|jjdddi|}t |j|||}|S) a*Moment test for binned probabilites using OPG. Paramters --------- binedges : array_like or None This defines which counts are included in the test on frequencies and how counts are combined in bins. The default if bin_edges is None will change in future. See Notes and Example sections below. method : str Currently only `method = "opg"` is available. If method is None, the OPG will be used, but the default might change in future versions. See Notes section below. Returns ------- test result Notes ----- Warning: The current default can have many empty or nearly empty bins. The default number of bins is given by max(endog). Currently it is recommended to limit the number of bins explicitly, see Examples below. Binning will change in future and automatic binning will be added. Currently only the outer product of gradient, OPG, method is implemented. In many case, the OPG version of a specification test overrejects in small samples. Specialized tests that use observed or expected information matrix often have better small sample properties. The default method will change if better methods are added. Examples -------- The following call is a test for the probability of zeros `test_chisquare_prob(bin_edges=np.arange(3))` `test_chisquare_prob(bin_edges=np.arange(10))` tests the hypothesis that the frequencies for counts up to 7 correspond to the estimated Poisson distributions. In this case, edges are 0, ..., 9 which defines 9 bins for counts 0 to 8. The last bin is dropped, so the joint test hypothesis is that the observed aggregated frequencies for counts 0 to 7 correspond to the model prediction for those frequencies. Predicted probabilites Prob(y_i = k | x) are aggregated over observations ``i``. Nrrrr) bin_edgesmethodr)rrrrr )rr"r#rprobsress rr z#CountDiagnostic.test_chisquare_prob7spd  !y2):;;D $ $::6:T::!$,)/111 r predictedc |jd}t|}tj|jjjt|d|}t|||||}|S)zFPlot observed versus predicted frequencies for entire sample. r) minlengthN)labelupp_xlimfig) rsumlenrbincountrmodelendogastypeintr )rr)r*r+rk_probsfreqs rr zCountDiagnostic.plot_probsrs.22155o&&{4<-3::3??%,....6wh8$x """ rr)NN)r&NN) __name__ __module__ __qualname____doc__rrrr r rrrr r s~  ::^:9999v6:      rr c4eZdZdZdZdZd dZ d d ZdS)PoissonDiagnosticzDiagnostic and specification tests and plots for Poisson model status: experimental Parameters ---------- results : PoissonResults instance c||_dSr)r)rrs r_init__zPoissonDiagnostic._init__s  rc.t|j}|S)z{Test for excess (over or under) dispersion in Poisson. Returns ------- dispersion results )rr)rr%s rtest_dispersionz!PoissonDiagnostic.test_dispersions&dl33 rrNc$|dkr+|tjdt|j}n^|dkrX|t |j}nAt j|}|jdkr |dddf}t|j|}|S)aTest for excess zeros, zero inflation or deflation. Parameters ---------- method : str Three methods ara available for the test: - "prob" : moment test for the probability of zeros - "broek" : score test against zero inflation with or without explanatory variables for inflation exog_infl : array_like or None Optional explanatory variables under the alternative of zero inflation, or deflation. Only used if method is "broek". Returns ------- results Notes ----- If method = "prob", then the moment test of He et al 1_ is used based on the explicit formula in Tang and Tang 2_. If method = "broek" and exog_infl is None, then the test by Van den Broek 3_ is used. This is a score test against and alternative of constant zero inflation or deflation. If method = "broek" and exog_infl is provided, then the extension of the broek test to varying zero inflation or deflation by Jansakul and Hinde is used. Warning: The Broek and the Jansakul and Hinde tests are not numerically stable when the probability of zeros in Poisson is small, i.e. if the conditional means of the estimated Poisson distribution are large. In these cases, p-values will not be accurate. rNz*exog_infl is only used if method = "broek"broekr) exog_infl) warningswarnrrrrasarrayndimr)rr#rAr%s rtest_poisson_zeroinflationz,PoissonDiagnostic.test_poisson_zeroinflationsL V  $ JKKK$T\22CC w   6t|DDJy11 >Q&& )!!!T' 2I3DL>GIII r quicksort皙?皙?c ||jd}|jjj}|jd} |dddft j| jdkt} |Xt|} | | d dz } t j | | |zkdz }| ddd|f} | ddd|f} | dddfxxd| dz z cc<| dddfxxd| dz z cc<t| | |||d|| } | S) aHosmer-Lemeshow style test for count data. Note, this does not take into account that parameters are estimated. The distribution of the test statistic is only an approximation. This corresponds to the Hosmer-Lemeshow type test for an ordinal response variable. The outcome space y = k is partitioned into bins and treated as ordinal variable. The observations are split into approximately equal sized groups of observations sorted according the ``sort_var``. Nlin)rrrrT)sort_varbinsdfordered sort_methodalpha_nc)rrr/r0rrshaper1r2r-r,cumsumargmaxr)rrNrOk_maxrPrRfrac_upprSr0expectedcountsnobsicumcounts_sumr%s r_chisquare_binnedz#PoissonDiagnostic._chisquare_binneds  |++%+88H "( <''f'554.BIhnQ.?$@$@@HHMM =u::D!FJJqMM$8$8$;$;;NInth>??!CEAAAvvI&6E6 " B1x||A..qqq"u VZZ]]** %VX*.2t1<.6888 r)rN)NrGNNrHrIrJ)r5r6r7r8r<r>rFr]rrrr:r:sr4444lHLrbs777777````````Fzzzzzzzzzzr