M/PhloLdZddlmZddlmZmZmZmZmZddl m Z ddl Z ddl mZmZmZddlmZddlmZdd lmZmZdd lmZmZdd lmZmZmZd Ze d gdZ d!e"eej#$dZ%dZ&dZ'dZ(e(Z)e&e'ze)zZ*dZ+edde%e*ee+GddZ,e&e)zZ*edde%e*ee+Gdde,Z-GddZ.dS)z Rolling OLS and WLS Implements an efficient rolling estimator that avoids repeated matrix multiplication. Copyright (c) 2019 Kevin Sheppard License: 3-clause BSD )lstsq)Appender Substitutioncache_readonlycall_cached_funcget_cached_doc) namedtupleN) DataFrame MultiIndexSeries)stats)model)LikelihoodModelResultsModel)RegressionModelRegressionResults) array_likeint_like string_likecD|dr |ddS|S)N ) startswith)lines ^/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/statsmodels/regression/rolling.pystrip4r"s( sABBx K RollingStore paramsssrllfnobss2xpxixeex centered_tssuncentered_tss zvwindow : int Length of the rolling window. Must be strictly larger than the number of variables in the model. z weights : array_like, optional A 1d array of weights. If you supply 1/W then the variables are pre- multiplied by 1/sqrt(W). If no weights are supplied the default value is 1 and WLS results are the same as OLS. a)min_nobs : {int, None} Minimum number of observations required to estimate a model when data are missing. If None, the minimum depends on the number of regressors in the model. Must be smaller than window. missing : str, default "drop" Available options are "drop", "skip" and "raise". If "drop", any observations with nans are dropped and the estimates are computed using only the non-missing values in each window. If 'skip' blocks containing missing values are skipped and the corresponding results contains NaN. If 'raise', an error is raised. Default is 'drop'. expanding : bool, default False If True, then the initial observations after min_nobs are filled using an expanding scheme until ``window`` observations are available, after which rolling is used. aB Rolling %(model_type)s Least Squares %(parameters)s %(extra_parameters)s See Also -------- statsmodels.regression.linear_model.%(model)s %(model)s estimation and parameter testing. Notes ----- Tested against %(model)s for accuracy. Results may differ from %(model)s applied to windows of data if this model contains an implicit constant (i.e., includes dummies for all categories) rather than an explicit constant (e.g., a column of 1s). Examples -------- >>> from statsmodels.regression.rolling import Rolling%(model)s >>> from statsmodels.datasets import longley >>> data = longley.load() >>> exog = add_constant(data.exog, prepend=False) >>> mod = Rolling%(model)s(data.endog, exog) >>> rolling_res = mod.fit(reset=50) Use params_only to skip all calculations except parameter estimation >>> rolling_params = mod.fit(params_only=True) Use expanding and min_nobs to fill the initial results using an expanding scheme until window observation, and the roll. >>> mod = Rolling%(model)s(data.endog, exog, window=60, min_nobs=12, ... expanding=True) >>> rolling_res = mod.fit() WeightedWLS) model_typer parametersextra_parametersceZdZ dddddddZdZdZdZd Zd Zd Z d Z ddZ e e ejj ddZdS) RollingWLSNdropFweightsmin_nobsmissing expandingcPt|dd}|dkrdnd}tj||||d|j} |jj} tj|||dd| |_| |j_t |d |_|jjd } t |d d | df |_ t|d d}t |dd| f}||n|jjd |_ |du|_ |tj| n||_tj|j} | |jz|_| dddf|j z|_t|dd}||n|j jd|_|j|j jdks|j|j krt)d||_tj|jt.|_||_|jj|_|dk|_dS)Nr5)r1raiseskipoptionsr8r1)r5hasconstnoneFendogr)ndimshapewindowToptionalr3)rDrAr4zWmin_nobs must be larger than the number of regressors in the model and less than windowdtyper9)rr__init__ k_constantdata const_idxr_yrA_xr_window _weightednpones_weightssqrt_wy_wx _min_nobs ValueError _expanding zeros_likebool_is_nan _find_nans_has_nan _skip_missing) selfr>exogrBr3r4r5r6 temp_msngk_constrKr#w12s rrHzRollingWLS.__init__s! Y(A   &00FFg  tUD)dKKKK/I'  tUD&5IIII!' UG,,w}QT74,GGG&(T:::Wi$tgNNN!'!3vvq9I  ,)0 g gdm$$=qqq$w<$')Hj4@@@%-%9tw}Q?O >DGM!, , ,0M0M?  $}TWD999 )) ,$.rc .tj|||||fi|SN)r _handle_data)r_r>r`r5r<kwargss rrfzRollingWLS._handle_datas/! %w  4:   rctj|j}|tjtj|jdz}|tj|jz}||jdd<tj|}|j}||dz d|d|dz z ||dz d<|j r d|d|j <n d|d|dz <| tS)NrEaxisF) rPisnanrLanyrMrRr[cumsumrNrXrVastyperZ)r_nanshas_nanws rr\zRollingWLS._find_nanssx   rx((q1111 ''' QQQ)D// L"1q577+gj!a%j.AAA ? %(-G$dn$ % %$GGa!eG ~~d###rc`|j}||krt||z |}nt|}|j|}|j|}|j|}|j|}|j|}|} tj|r || }|| }|| }|| }||||| fSre) rNslicerLrTrUrRr[rPrl) r_idxrBlocywywxr3r5 not_missings r _get_datazRollingWLS._get_datas &==f c**CC**C GCL Xc] Xc]-$,s#h 6'?? ++AKBKBk*G"b';..rcj||jkrdS |dks|stj|}|dkr||z} n\||\} } } } } |dkrt | | d} n$tj| } | | z} n#tjj$rYdSwxYw| |j|dz <|rdS||\}} } }} | | | | ||\}}}| |dddfz}|j |z}| j d}|||z z }| || |\}}||j |dz <||j|dz <||j|dz <||j|dz <||j|dz <||j|dz <||j|dz <||j|dz <dS)NinvrrrE)rVrPlinalgr|rzrpinv LinAlgErrorr _loglikeTrA_sum_of_squaresr!r"r#r$r%r&r'r()r_rtwxpwxwxpwyr#store params_onlymethodwxpwxir _rwrx wxpwxiwxprvr3wresidr!r"wxwresidwxepwxe tot_paramsr$r'r(s r _fit_singlezRollingWLS._fit_singles $. F % u--%"&.."5"52r1aW$$"2r]]1-FF " r 2 2I&^Fy$    FF  & S1W   F $s 3 32r7A==R$GGSqqq$w'*x'Xa[ D:% &'+';';Ar7'K'K$ n  #'  #'" 37q$ 37% 37&237#(6S1W%%%sBBB76B7c4|dz }|||zz }tj|dzd}tj| |z} | dtjtj|z z|zz} | dtjtj|zz } ||| fS)Ng@r?rrirEg?)rPsumlogpi) r_r rwrxr3r#nobs2rr!r"s rrzRollingWLS._loglike ss b6k!fVq[q)))vc{{lU" BF255=)))U22 sRVBF7OO,,,,sCrctj||}tj|||z dzz}tj||}||fS)N)r3r?)rPaveragerdot)r_rvrwr3meanr'r(s rrzRollingWLS._sum_of_squaressLz!W---vgTa788 B^++rc||\}}}}}|}|j|z}|j|z}|||fS)z.Compute xpx and xpy using a single dot product)rzrr) r_rtrrwrxryr#xpxxpys r_resetzRollingWLS._resetsO$(NN3$7$7!2r1k  dRidRiC~rr| nonrobustcTt|dd}t|dd}||jjdn|}|d krt d |jj\}}t tj||ftj tj|tj tj|tj tj |t tj|tj tj|||ftj tj|||ftj tj|tj tj|tj  } |j } |j r|jn| } || \} } }|j| d z r|js|| | | || |||j|j}}t+| d z|jjdd zD]}|j|d z r|jr||zdkr||\} } }n|j|| z d z sI|| krC||| z d z || z }| |j|zz} | |j||| z d z || z zz} |d z}|j|d z s7||d z |}| |j|zz } | |j||d z |zz } |d z }||| | || ||t1|| |j||S) a Estimate model parameters. Parameters ---------- method : {'inv', 'lstsq', 'pinv'} Method to use when computing the the model parameters. * 'inv' - use moving windows inner-products and matrix inversion. This method is the fastest, but may be less accurate than the other methods. * 'lstsq' - Use numpy.linalg.lstsq * 'pinv' - Use numpy.linalg.pinv. This method matches the default estimator in non-moving regression estimators. cov_type : {'nonrobust', 'HCCM', 'HC0'} Covariance estimator: * nonrobust - The classic OLS covariance estimator * HCCM, HC0 - White heteroskedasticity robust covariance cov_kwds : dict Unused reset : int, optional Interval to recompute the moving window inner products used to estimate the model parameters. Smaller values improve accuracy, although in practice this setting is not required to be set. use_t : bool, optional Flag indicating to use the Student's t distribution when computing p-values. params_only : bool, optional Flag indicating that only parameters should be computed. Avoids calculating all other statistics or performing inference. Returns ------- RollingRegressionResults Estimation results where all pre-sample values are nan-filled. r)r|rr~r:resetTrCNrrEz reset must be a positive integerrFr)rrrLrArWrMrrPfullnanzerosintrNrXrVrr]r^rrUrTranger[rRollingRegressionResultsrI)r_rcov_typecov_kwdsruse_trr#krrqfirstrrrxrwiremove_xadd_xs rfitzRollingWLS.fit#s7\ H&>   $777$)M a  u 199?@@ @'-a7D!9bf--bf%%bf%%$c***wtRV$$$1rv..$1rv..rv..7400     L"&/8qU++S$ eai( PT-? P   UCdE; O O O48Buqy$'-"2Q"677 L LA}QU# (: 5yA~~!%QS$$|AEAI.1q55!!a%!)a!e"34H8:00C8:1q519q1u+<(===CAID|AE*q1uqyME57U?*C57RA ]22CAID   QS${F K K K K' %%   rc.| |j|}|dd}|d}n|dkrddlm} | i}n|dz }|dd} dd lm} m} | d g } | |||d | }|\}}|jdkr|jddks |jdkr'td |j| | |d|||d<|||g|Ri|}||_ ||j _|S)Neval_envr?r)EvalEnvironmentrEr5r9)NAAction dmatricesr8)on_NANA_types dataframe) return_type NA_actionzendog has evaluated to an array with multiple columns that has shape {}. This occurs when the variable converted to endog is non-numeric (e.g., bool or str).)r5rBr3)rupoppatsyrgetrrr@rArWformatupdateformularJframe)clsrrJrBr3subsetargsrgrrr5rr na_actionresultr>r`mods r from_formulazRollingWLS.from_formulas  8F#D::j$//  HH ^^ - - - - - -&r**HH MH**Y//--------H7R888    #     t JNNu{1~11ej1nn()/u{(;(;    'V<<===   'F9 c%//////  rre)r|rNNFF)NN)__name__ __module__ __qualname__rHrfr\rzrrrrr classmethodrrr__doc__rrr0r0s 0/ 0/0/0/0/0/d   $ $ $///&&7&7&7P   ,,, ] ] ] ] ~ Xe ())9=****)[***rr0OrdinaryOLSc,eZdZ dddddfd ZxZS) RollingOLSNr1F)r4r5r6c Xt|||d|||dS)Nr2)superrH)r_r>r`rBr4r5r6 __class__s rrHzRollingOLS.__init__sE          rre)rrrrH __classcell__)rs@rrrsW             rrcXeZdZdZeZdefdZdZe e e e j dZ e e e e jdZd*dZe d Ze e e e jd Ze e e e jd Ze e ejjd Ze d Ze e e e jdZe e e e jdZe e e e jdZe e e e jdZe e e e jdZe e ejjdZe e e jjdZe e e e jdZe e e e jdZe e e e jdZe e e e j dZ e dZ!dZ"e e e e j#dZ#e e e e j$dZ$e e e e j%dZ%e e e e&j'dZ'e e e e&j(dZ(d Z)e e&j*jd+d#Z*e+d$Z,e-e e&j.jd%Z.e&j/Z/e e&j0jd,d'Z0 d-d)Z1d"S).ra Results from rolling regressions Parameters ---------- model : RollingWLS Model instance store : RollingStore Container for raw moving window results k_constant : bool Flag indicating that the model contains a constant use_t : bool Flag indicating to use the Student's t distribution when computing p-values. cov_type : str Name of covariance estimator rc||_|j|_|j|_|j|_|j|_|j |_ |j |_ |j |_|j|_|j|_||_|j jd|_||dk}||_||_|jjjdu|_g|_i|_dS)Nrr)rr _paramsr!_ssrr"_llfr#_nobsr$_s2r%_xpxir&_xepxer' _centered_tssr(_uncentered_tss _k_constantrA_nvar_use_t _cov_typerJ row_labels _use_pandas _data_attr_cache)r_rrrIrrs rrHz!RollingRegressionResults.__init__s | I I Z 8Z j "/$3%Z%b) = +E !:?5TA rc^|js|S|jjj}|jjj}|jdkrt ||S|jdkrt|||Stj ||f}tj |d|j df}t|||S)z4Wrap output as pandas Series or DataFrames as neededrE)indexr?columnsrr) rrrJ param_namesrr@r r r from_productrPreshaperA)r_val col_names row_namesmis r_wrapzRollingRegressionResults._wraps JJO/ JO. 8q==#Y/// / 8q==S)9EEE E()Y)?@@B*S2sy}"566CS)2>>> >rc\|ttj|Sre)rrraicr_s rrzRollingRegressionResults.aic %zz*+<+@$GGHHHrctjd5|ttj|cdddS#1swxYwYdS)Nignore)divide)rPerrstaterrrbicrs rrzRollingRegressionResults.bics[ ) ) ) M M::./@/DdKKLL M M M M M M M M M M M M M M M M M M-AAArcV|tj|||S)N) dk_params)rr info_criteria)r_critrs rrz&RollingRegressionResults.info_criterias-zz  +D$) L L L   rc6||jS)zEstimated model parameters)rrrs rr zRollingRegressionResults.paramsszz$,'''rc6||jSre)rrrs rr!zRollingRegressionResults.ssr!zz$)$$$rc6||jSre)rrrs rr"zRollingRegressionResults.llf&rrc |j|jz Sre)rrrs rdf_modelz!RollingRegressionResults.df_model+szD,,,rc|jS)z5Flag indicating whether the model contains a constant)rrs rrIz#RollingRegressionResults.k_constant0s rc|jSre)rrs rr'z%RollingRegressionResults.centered_tss5s !!rc|jSre)rrs rr(z'RollingRegressionResults.uncentered_tss:s ##rc\|ttj|Sre)rrrrsquaredrs rr z!RollingRegressionResults.rsquared?s%zz*+<+EtLLMMMrc\|ttj|Sre)rrr rsquared_adjrs rrz%RollingRegressionResults.rsquared_adjDs+zz .;T B B   rc6||jSre)rrrs rr#zRollingRegressionResults.nobsKszz$*%%%rcV||j|jz |jz Sre)rrrrrs rdf_residz!RollingRegressionResults.df_residPs'zz$*t}4t7GGHHHrc|jSre)rrs rrzRollingRegressionResults.use_tUs {rc\|ttj|Sre)rrressrs rrzRollingRegressionResults.essZrrc\|ttj|Sre)rrr mse_modelrs rrz"RollingRegressionResults.mse_model_%zz*+<+FMMNNNrc\|ttj|Sre)rrr mse_residrs rrz"RollingRegressionResults.mse_residdrrc\|ttj|Sre)rrr mse_totalrs rrz"RollingRegressionResults.mse_totalirrcz|jdkr|jddddf|jzS|j|jz|jzS)Nr)rrrrrs r _cov_paramsz$RollingRegressionResults._cov_paramsnsC >[ ( (8AAAtTM*TZ7 7: +dj8 8rc6||jS)a Estimated parameter covariance Returns ------- array_like The estimated model covariances. If the original input is a numpy array, the returned covariance is a 3-d array with shape (nobs, nvar, nvar). If the original inputs are pandas types, then the returned covariance is a DataFrame with a MultiIndex with key (observation, variable), so that the covariance for observation with index i is cov.loc[i]. )rrrs r cov_paramsz#RollingRegressionResults.cov_paramsuszz$*+++rctjd5|ttj|cdddS#1swxYwYdSNrinvalid)rPrrrrf_pvaluers rr%z!RollingRegressionResults.f_pvalues[ * * *  :: !2!;TBB                  rc|jdkr|j|jz S|jjd}t j|t j}|jjd}t j|}tt|}|j r| |j j|s|S||}|j}||z|jz}|j}t|D]n} || | dz|jz} | jd} t j|| } t j| | z| jz| z || <o|S)NrrrE)rrrrrArPrreyelistrrIrrrKrrr}r|squeeze) r_r#statrrlocsvcvrvcvrprrpdenominv_covs rfvaluezRollingRegressionResults.fvalues? >[ ( (>DN2 2<%a(D74((D "1%Aq Aa>>D /-...  $A"CGacME A4[[ B Bq1q5y\AC' )--a11*R'\BD%899EAQKrc tjd5|tjtj|jddcdddS#1swxYwYdS)Nrr#rr?)rPrrrSdiagonalrrs rbsezRollingRegressionResults.bses[ * * * L L::bgbk$2BAq&I&IJJKK L L L L L L L L L L L L L L L L L LsAA##A'*A'ctjd5|ttj|cdddS#1swxYwYdSr")rPrrrrtvaluesrs rr8z RollingRegressionResults.tvaluess[ * * *  :: !7!?FF                  rc|jrt|d|j}tj|dddf}tjd5t jtj |j |dzcdddS#1swxYwYdStjd5t j tj |j dzcdddS#1swxYwYdS)Ndf_resid_inferencerr#r?) rgetattrrrPasarrayrr tsfabsr8norm)r_rs rpvaluesz RollingRegressionResults.pvaluess : ?t%94=IIHz(++AAAtG4HX... F Fwzz"&"6"6AAAE F F F F F F F F F F F F F F F F F FX... ? ?z}}RVDL%9%9::Q> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?s$:BBB:9DDDcVtj|j}|jr]tj}t |d|j}tj|dddf}|d|dz z |}n'tj }|d|dz z }tj|j }|||zz }|||zz} |,tj|}|dd|f}| dd|f} tjtt|| S)Nr:rEr?) rPr<r6rr r=r;rppfr@r r(zip) r_alphacolsr6distrqr loweruppers r _conf_intz"RollingRegressionResults._conf_intsj"" : (7Dt%94=IIHz(++AAAtG4HUQY11AA:DUQY''ADK((S S   :d##D!!!T'NE!!!T'NEz$s5%0011222r皙?Ncf|||}|js|Sd}|jjj}|jjj|fd|Dt j|f}tj tj |dd|j ddf}t|||S)N)rIrJc g|] }| Srr).0rrs r z5RollingRegressionResults.conf_int..s444!1444rrEr?rrr) rKrrrJrrr rrPrswapaxesrAr )r_rErFcici_namesrrrs @rconf_intz!RollingRegressionResults.conf_ints ^^E4 ( ( I%JO. JO/  4444t444I  $i%: ; ; Z B1-- R/@ A ARy9999rc|jS)zName of covariance estimator)rrs rrz!RollingRegressionResults.cov_types ~rc*tj|Sre)rload)rfnames rrWzRollingRegressionResults.loads&*5111rFc.tj|||Sre)rsave)r_rX remove_datas rrZzRollingRegressionResults.saves%*4 DDDr upper leftcddlm}m}|||d}|jjj} | $tj|j j d} |j j d} |jjj } |tt| } nt|tt fr|g}g} tt#|D]} || }|| vr)| | |7t|tr| |bd|| d| }t-|||||}d}ddl}t| |jr| } tj| } | D]}} |t#| d|dz}|j dd| f}tj|j dd| f}| |}|||||tj|dddd| fd}tjtj|sn|||dddf|dd |||dddf|dd |dkr| | |!|d|d |"| | |dz }|#|S)ai Plot the recursively estimated coefficients on a given variable Parameters ---------- variables : {int, str, Iterable[int], Iterable[str], None}, optional Integer index or string name of the variables whose coefficients to plot. Can also be an iterable of integers or strings. Default plots all coefficients. alpha : float, optional The confidence intervals for the coefficient are (1 - alpha)%. Set to None to exclude confidence intervals. legend_loc : str, optional The location of the legend in the plot. Default is upper left. fig : Figure, optional If given, subplots are created in this figure instead of in a new figure. Note that the grid will be created in the provided figure using `fig.add_subplot()`. figsize : tuple, optional If a figure is created, this argument allows specifying a size. The tuple is (width, height). Returns ------- Figure The matplotlib Figure object. r) _import_mplcreate_mpl_figNrEzQvariable {} is not an integer and was not found in the list of variable names: {}z, )rr?zk:zLower CI)labelzUpper CI)rur)$statsmodels.graphics.utilsr^r_rKrrJrrParangerrArr(r isinstancerstrlenappendrrjoinrWpandas PeriodIndex to_timestampr< add_subplotrkplotralllegendset_xlim set_title tight_layout)r_ variablesrE legend_locfigfigsizer^r_rRr k_variablesr variable_idxrvariablemsgrupdaxr validrow_lblthis_cis rplot_recursive_coefficientz3RollingRegressionResults.plot_recursive_coefficientsF KJJJJJJJ  t,,BZ_/  4<#5a#899Jl(+ jo1  k 2 233LL)c3Z00 (&K L3y>>** * *$Q<{** '' (9(9((C(CDDDD#.. * ''1111$$*F%aL$))K*@*@%%%S//) nS'** j". 1 1 3#0022JZ ++   A\!2!2AsQw??B\!!!Q$'FXdl111a40111E 'G GGGVE] + + + *R111a['::vbhw//002GGAu!5t:GGAu!5t:axx j 111 KK GBK 0 0 0 LLQ ( ( ( 1HCC  r)r)rLN)F)NrLr\NN)2rrrrtuple_data_in_cacherrHrrrrrrrrr r!r"rrrIr'r(r rr#rrrrrrrr r%r3r6rr8rArKrTpropertyrrrWr[rZrrrrrrs$UWWN(. ? ? ? Xnn.23344II54^I Xnn.23344MM54^M    ((^( Xnn.23344%%54^% Xnn.23344%%54^% Xo&.//--0/^-  ^  Xnn.;<<=="">=^" Xnn.=>>??$$@?^$ Xnn.78899NN:9^N Xnn.;<<==  >=^  Xnn.34455&&65^& Xo&.//II0/^I X%-../.^ Xnn.23344II54^I Xnn.899::OO;:^O Xnn.899::OO;:^O Xnn.899::OO;:^O99^9 ,,,  Xnn.78899:9^  Xnn.5667787^0 Xnn.23344LL54^L Xnn3;<<==>=^  Xnn3;<<==??>=^?333*X$-566 : : :76 :X X$)1222232[2)4K X$)122EEE32E   bbbbbbrr)/rstatsmodels.compat.numpyrstatsmodels.compat.pandasrrrrr collectionsr numpyrPrhr r r scipyr statsmodels.baserstatsmodels.base.modelrr#statsmodels.regression.linear_modelrrstatsmodels.tools.validationrrrrrrgmap_model_params_docsplit common_paramswindow_parametersweight_parameters_missing_param_doc extra_baser._docr0rrrrrrs+*****#"""""0000000000""""""@@@@@@@@KJJJJJJJJJ z       ##fe&=&C&CD&I&IJJKK  $ $'88:E&R %   $ddddddd dN %z1 %   $         ,EEEEEEEEEEr