%gA2,ddlmZmZmZddlZddlZddlmZmZddl m Z ddl m Z ddl mZmZmZddlZddlZddlmZddlmZmZdd lmZmZmZmZdd lm Z m!Z!ejDd Z#e#jIejJd Z&Gd de'Z(y))absolute_importdivisionprint_functionN) OrderedDict defaultdict)deepcopy) timedelta)DictListUnion)NDArray)get_holiday_namesmake_holidays_df)StanBackendEnumModelInputData ModelParamsTrendIndicator)plotplot_componentsprophetiʚ;ceZdZdZ dIdefdZdZdZ dJdZdKdZ d Z d Z e d e jd eeefd edeej(fdZedZdZdZ dLdZ dMdZdZdZdZdZdZ dZ!e dZ"e dZ#e dZ$de jJde&fdZ'd ede(fd!Z)d"Z*dNde jJd#e+de jJfd$Z,e d%Z-e d&Z.e d'Z/d(Z0d)Z1de jJd#e+de jJfd*Z2de jJd#e+de3eejhffd+Z5de3eejhffd,Z6de jJd-e jJd.ed/ejhd0ejhd1ede7e3eejhffd2Z8d3Z9dOde jJd1ed.edejhfd4Z:dOde jJd1ed.edejhfd5Z;e d6ed7ed8ed1edejhf d9Zed?ejhd@ejhdAedBedejhfdCZ>dRde jJd#e+fdDZ?dEZ@dSdFZA dTdGZB dUdHZCy)VPropheta Prophet forecaster. Parameters ---------- growth: String 'linear', 'logistic' or 'flat' to specify a linear, logistic or flat trend. changepoints: List of dates at which to include potential changepoints. If not specified, potential changepoints are selected automatically. n_changepoints: Number of potential changepoints to include. Not used if input `changepoints` is supplied. If `changepoints` is not supplied, then n_changepoints potential changepoints are selected uniformly from the first `changepoint_range` proportion of the history. changepoint_range: Proportion of history in which trend changepoints will be estimated. Defaults to 0.8 for the first 80%. Not used if `changepoints` is specified. yearly_seasonality: Fit yearly seasonality. Can be 'auto', True, False, or a number of Fourier terms to generate. weekly_seasonality: Fit weekly seasonality. Can be 'auto', True, False, or a number of Fourier terms to generate. daily_seasonality: Fit daily seasonality. Can be 'auto', True, False, or a number of Fourier terms to generate. holidays: pd.DataFrame with columns holiday (string) and ds (date type) and optionally columns lower_window and upper_window which specify a range of days around the date to be included as holidays. lower_window=-2 will include 2 days prior to the date as holidays. Also optionally can have a column prior_scale specifying the prior scale for that holiday. seasonality_mode: 'additive' (default) or 'multiplicative'. seasonality_prior_scale: Parameter modulating the strength of the seasonality model. Larger values allow the model to fit larger seasonal fluctuations, smaller values dampen the seasonality. Can be specified for individual seasonalities using add_seasonality. holidays_prior_scale: Parameter modulating the strength of the holiday components model, unless overridden in the holidays input. changepoint_prior_scale: Parameter modulating the flexibility of the automatic changepoint selection. Large values will allow many changepoints, small values will allow few changepoints. mcmc_samples: Integer, if greater than 0, will do full Bayesian inference with the specified number of MCMC samples. If 0, will do MAP estimation. interval_width: Float, width of the uncertainty intervals provided for the forecast. If mcmc_samples=0, this will be only the uncertainty in the trend using the MAP estimate of the extrapolated generative model. If mcmc.samples>0, this will be integrated over all model parameters, which will include uncertainty in seasonality. uncertainty_samples: Number of simulated draws used to estimate uncertainty intervals. Settings this value to 0 or False will disable uncertainty estimation and speed up the calculation. stan_backend: str as defined in StanBackendEnum default: None - will try to iterate over all available backends and find the working one holidays_mode: 'additive' or 'multiplicative'. Defaults to seasonality_mode. Nscalingc||_||_|j[tjtj|jd|_t |j|_d|_n||_d|_||_||_ ||_ ||_ ||_ | |_ ||_||j|_t| |_t| |_t| |_| |_||_||_|dvr t-d||_d|_d|_d|_d|_d|_d|_t=i|_t=i|_ d|_!d|_"i|_#d|_$d|_%d|_&d|_'d|_(i|_)|jU|jW|y)NdsnameTF)absmaxminmaxz+scaling must be one of 'absmax' or 'minmax'),growth changepointspdSeries to_datetimelenn_changepointsspecified_changepointschangepoint_rangeyearly_seasonalityweekly_seasonalitydaily_seasonalityholidaysseasonality_mode holidays_modefloatseasonality_prior_scalechangepoint_prior_scaleholidays_prior_scale mcmc_samplesinterval_widthuncertainty_samples ValueErrorrstarty_miny_scalelogistic_floort_scalechangepoints_tr seasonalitiesextra_regressorscountry_holidaysstan_fitparamshistory history_datestrain_component_colscomponent_modestrain_holiday_names fit_kwargsvalidate_inputs_load_stan_backend)selfr r!r&r(r)r*r+r,r-r0r2r1r3r4r5 stan_backendrr.s O/var/www/dash_apps/app1/venv/lib/python3.12/site-packages/prophet/forecaster.py__init__zProphet.__init__Rs* (    ( " "..9J9J*KRV WD "%d&7&7"8D *.D '"0D */D '!2"4"4!2  0*  !%!6!6D ',-D'E$',-D'E$$)*>$?!(,#6 . .JK K    # "(_ +B $   !$(!##'    -c|ItD]?} tjd|j|j |jcSntj ||_tjd|jjy#t $r+}tjd|j|Yd}~d}~wwxYw)NzTrying to load backend: %sz3Unable to load backend %s (%s), trying the next onezLoaded stan backend: %s) rloggerdebugrrI Exceptionget_backend_classrKget_type)rJrKies rLrIzProphet._load_stan_backends  $ ccLL!=qvvF22166:: c!P A A, O QD  .0A0A0J0J0LM !cLL!VXYX^X^`abbcs:B C#!C  Cc|jdvr tdt|jtt fs td|jdks|jdkDr td|j t|j tjrd|j vrd |j vs td tj|j d|j d<|j djjs+|j d jjr td d |j v}d |j v}||zdk(r td|rV|j d jdkDr td|j d jdkr td|j d jD]}|j|d|j dvr td|j"dvr tdy)z Validates the inputs to Prophet.)linearlogisticflatz= 0Fcheck_holidaysadditivemultiplicativez7seasonality_mode must be "additive" or "multiplicative"z4holidays_mode must be "additive" or "multiplicative")r r6 isinstancer(intr/r,r" DataFramer$isnullanymaxminuniquevalidate_column_namer-r.)rJ has_lower has_upperhs rLrHzProphet.validate_inputss ;;< <NP P$003,?LM M  # #a 'T-C-Ca-GNO O == $4==",,7DMM). "677"$..t1D"EDMM$  d#**,002==+22488: !EFF&$--7I&$--7I9$) "<====0446:$%JKK==0446:$%JKK]]9-446 C))!E)B C  (F FI    %C CF  DrNcd|vr tdgd}|Dcgc]}|dz }}|Dcgc]}|dz }}|j||j||jgd||vrtdj||rF|j:||jd j vrtd j||rI|j =|t |j vr&td j||j |r)||jvrtdj||r*||jvrtdj|y y cc}wcc}w)aYValidates the name of a seasonality, holiday, or regressor. Parameters ---------- name: string check_holidays: bool check if name already used for holiday check_seasonalities: bool check if name already used for seasonality check_regressors: bool check if name already used for regressor _delim_zName cannot contain "_delim_") trendadditive_termsdailyweeklyyearlyr,zerosextra_regressors_additiveyhatextra_regressors_multiplicativemultiplicative_terms_lower_upper)rycapfloory_scaled cap_scaledzName {name!r} is reserved.rNr\z)Name {name!r} already used for a holiday.z6Name {name!r} is a holiday name in {country_holidays}.)rr?z-Name {name!r} already used for a seasonality.z2Name {name!r} already used for an added regressor.) r6extendformatr,rkr?rr=r>) rJrr`check_seasonalitiescheck_regressorsreserved_namesnrn_lrn_us rLrlzProphet.validate_column_names  <= =  '55H 55&45H 55d#d#A B > !,333>  t}}8 i07799;BBBM  t44@)$*?*?@@HTD4I4IJ  44+=+=#=?T"  (=(= =DT" !> 565s E7 E<cRd|vrVtj|d|d<tj|djj r t d|djtjk(r|djt|d<tj|d|d<|djj t d|djj r t d|jD]x}||vrt dj!|tj||||<||jj s_t d j!||j"j D]}|d }| ||vrt d j!| ||j%d dgj'st dj!| ||jd||<|j(j*dk(rd|j(_|j-dd}|j/d }|j1|||j2rd|vr>t d|j4dk(rd|d<n|j4dk(r|j6|d<|j8dk(rNd|vr t d|d|dkj r t d|d|dz |j:z |d<|d|j<z |j>z |d<d|vr|d|dz |j:z |d<|jjAD]\}}|||d z |d!z ||<|S)"aFPrepare dataframe for fitting or predicting. Adds a time index and scales y. Creates auxiliary columns 't', 't_ix', 'y_scaled', and 'cap_scaled'. These columns are used during both fitting and predicting. Parameters ---------- df: pd.DataFrame with columns ds, y, and cap if logistic growth. Any specified additional regressors must also be present. initialize_scales: Boolean set scaling factors in self from df. Returns ------- pd.DataFrame prepared for fitting or predicting. r~zFound infinity in column y.rNzJColumn ds has timezone specified, which is not supported. Remove timezone.zFound NaN in column ds.z)Regressor {name!r} missing from dataframerzFound NaN in column {name!r}condition_namez3Condition {condition_name!r} missing from dataframe)rTFz.Found non-boolean in column {condition_name!r}bool mergesort)kind)droprzExpected column "floor".rrrYrz?Capacities must be supplied for logistic growth in column "cap"z5cap must be greater than floor (which defaults to 0).rtrmustd)!r" to_numericnpisinfvaluesrhr6dtypeint64astypestrr$dttzrgr>rr=isinallindexr sort_values reset_indexinitialize_scalesr:rr8r r9r7r;items)rJdfrrpropsrs rLsetup_dataframezProphet.setup_dataframes" "9mmBsG,BsGxx3'++- !>?? d8>>RXX %$xs+BtH>>"T(+4 d8;;>> %#  d8??  "67 7)) D2~ ?VV&}}RX.BtH$x $$& 299t9D ''..0 GE"#34N)!+$M~>.)..e}=AAC$H~>&(%7%>%>v%F>" G 88==D  BHHM ^^D{^ 3 ^^^ & 0"5   b  !;<<||x' 7 )"jj7 ;;* $B #5 R[(--/ K!#5 BwK 74<<GB| d8djj(DLL83 "9 g7 3t||CBzN00668 AKD%DE$K/5<?BtH A rNcb|sy|jdk(rd|vrd|_|jdk(rkt|d|dz j j |_t|d|dz j j|_n|jdk(r|dj |_t|dj|j z |_n|jdk(r7d |_t|dj j|_nU|jdk(rF|dj |_t|dj|j z |_|jd k(rd |_|d j |_ |d j|jz |_ |jjD]\}}|d }t||j}|dkrd}|dk(r&t||jdd hk(rd}nd}|s\t||j!}t||j#}||j|d<||j|d<y)zInitialize model scales. Sets model scaling factors using df. Parameters ---------- initialize_scales: Boolean set the scales or not. df: pd.DataFrame for setting scales. NrYrTrr~rrrr?r standardizeFautor[rr)r r:rr/absrjr8rir9r7r;r>rr%rksetmeanr) rJrrrrrn_valsrrs rLrzProphet.initialize_scalesds?!  ;;* $B"&D ||x'"BsGbk$9#>#>#@#D#D#FG $bg7 &;%@%@%B%F%F%HI )[__. $RY]]_tzz%AB ||x' $bg]]_%8%8%:; )W[[] %bgkkmdjj&@ A <<1 DLX\\^ $x||~ 2 00668 9KD% .KD*+Fz# f$r$x()aV3"'K"&K2d8==?+BtHLLN+46%%d+D158%%d+E2 9rNc|jt|jdk(rnt|j|jdjk}t |j|jdj kD}|s|rOt dt tj|jjd|jz}|jdz|kDr9|dz |_ tjdj|j|jdkDrytjd|dz |jdzj!j#t }|jj$|dj'd|_n/t)j*t)j,gd |_t|jdkDrRtj.tj0|j|j2z |j4z |_ytj0dg|_y) aLSet changepoints Sets m$changepoints to the dates of changepoints. Either: 1) The changepoints were passed in explicitly. A) They are empty. B) They are not empty, and need validation. 2) We are generating a grid of them. 3) The user prefers no changepoints be used. Nrrz,Changepoints must fall within training data.r[zKn_changepoints greater than number of observations. Using {n_changepoints}.)r&r)r!r%rjrBrir6rerrshaper(r&rPinforlinspaceroundriloctailr"r#r$sortarrayr7r;r<)rJtoo_lowtoo_high hist_size cp_indexess rLset_changepointszProphet.set_changepointss    (4$$%*d//04<<3E3I3I3KKt001DLL4F4J4J4LLh$FHH BHHT\\%7%7%:'+'='=&>?@I""Q&2&/!m# .V4+>+>V? ""Q&KK9q=$2E2E2IJ LL%%j1$7<= 1)rg@rrN) r6to_numpyrrNANOSECONDS_TO_SECONDSpiemptyrrangesincos)rrrrx_Tfourier_componentsrUcs rLfourier_serieszProphet.fourier_seriess&!89 9 NNN *.D D S"%%i!mXXu{{1~q<7G&HI|$ ;Aq1u &A+-66!9 q!a%x (13 q1q5A+~ . ;"!rNc|j|||}t|jdDcgc]}dj||dz}}t j ||Scc}w)apData frame with seasonality features. Parameters ---------- cls: Prophet class. dates: pd.Series containing timestamps. period: Number of days of the period. series_order: Number of components. prefix: Column name prefix. Returns ------- pd.DataFrame with seasonality features. r[z {}_delim_{})columns)rrrrr"rf)clsrrrprefixfeaturesrUrs rLmake_seasonality_featuresz!Prophet.make_seasonality_featuressn %%eV\B8>>!,-   Q /  ||Hg66  sA#c 8tj}|j|jj}|jft |Dchc]}|j c}}t||j}tj||fd}|jdd|j|jtj|jj|j}|j!|}tjd|jtj|jj|ji}tj||fd}|jdd|Scc}w)aConstruct a dataframe of holiday dates. Will combine self.holidays with the built-in country holidays corresponding to input dates, if self.country_holidays is set. Parameters ---------- dates: pd.Series containing timestamps used for computing seasonality. Returns ------- dataframe of holiday dates, in holiday dataframe format used in initialization. ) year_listcountryF)rT)rinplacer\)r"rfr,copyr?listyearrconcatrrFrr logical_notr\rr)rJr all_holidaysxrcountry_holidays_df index_to_dropholidays_to_adds rLconstruct_holiday_dataframez#Prophet.construct_holiday_dataframesh||~ == $==--/L  ,e4aff45I"2#T-B-B# 99l4G%H*/1L  $ $$ $ =  # # /(.. ((--d.F.FGM (,,];L ll433NN4#;#;(,\-A-A(BD,O 99lO%D*/1L  $ $$ $ =55sFc Htfd}i}tjjj}|j D]}}|j j } tt|dd}tt|dd} tt|d|j} tj| rt|j} |j|vr7||j| k7r%tdj|j| dkr td| ||j<t!|| d zD]s} tj"|t%| z} |j'| } d j|j| dk\rd ndt+| }| d||| <o||utj,|}|t/|j0j3}|j0Dcgc]}||j5dd}}t7|j9}|j:tj<||_|||fS#t$rd}d} YwxYw#t($rd } YwxYwcc}w)aConstruct a dataframe of holiday features. Parameters ---------- dates: pd.Series containing timestamps used for computing seasonality. holidays: pd.Dataframe containing holidays, as returned by construct_holiday_dataframe. Returns ------- holiday_features: pd.DataFrame with a column for each holiday. prior_scale_list: List of prior scales for each holiday column. holiday_names: List of names of holidays cHtjjdS)Nr)rrwr)rsrLz/Prophet.make_holiday_features..=sQ0HrNr]rr^ prior_scalezGHoliday {holiday!r} does not have consistent prior scale specification.)r\Prior scale must be > 0r[daysNz {}_delim_{}{}+-rrq)rr" DatetimeIndexrdate itertuplesrregetattrr6r/r2risnanr\rrr$r get_locKeyErrorrrfsortedrtolistsplitrkeysrFr#)rJrr,expanded_holidays prior_scales row_indexrowrlwuwpsoffset occurrencelockeyholiday_featuresroprior_scale_list holiday_namess ` rLmake_holiday_featureszProphet.make_holiday_features-s ((HI $$UXX]]3 &&(" +CB na89na89wsM43L3LMNBxx|4445{{l*|CKK/HB/N ++16#++6+FQw !:;;(*L %BF+ +^^B1G,GH #++J7C&,,KK!Q;CCK ?24%c*3/%c* +)" +F<<(9:+F3C3K3KDJFH-NO&--  +A. /  \..01  # # +')yy'?D $!1=@@U  & C s**,I99J J9 J  J  JJc |j td|j|d|t|j}| |j }|dkr t d|dvr t d||dd |d |j|<|S) a[Add an additional regressor to be used for fitting and predicting. The dataframe passed to `fit` and `predict` will have a column with the specified name to be used as a regressor. When standardize='auto', the regressor will be standardized unless it is binary. The regression coefficient is given a prior with the specified scale parameter. Decreasing the prior scale will add additional regularization. If no prior scale is provided, self.holidays_prior_scale will be used. Mode can be specified as either 'additive' or 'multiplicative'. If not specified, self.seasonality_mode will be used. 'additive' means the effect of the regressor will be added to the trend, 'multiplicative' means it will multiply the trend. Parameters ---------- name: string name of the regressor. prior_scale: optional float scale for the normal prior. If not provided, self.holidays_prior_scale will be used. standardize: optional, specify whether this regressor will be standardized prior to fitting. Can be 'auto' (standardize if not binary), True, or False. mode: optional, 'additive' or 'multiplicative'. Defaults to self.seasonality_mode. Returns ------- The prophet object. z0Regressors must be added prior to model fitting.F)rrrraz+mode must be 'additive' or 'multiplicative'rr)rrrrmode)rBrRrlr/r2r-r6r>)rJrrrr s rL add_regressorzProphet.add_regressorts< << #BD D !!$!?   9 9:K <((D ! 67 7 5 5JK K&& ' d# rNc`|j td|dvr|j|d| |j}n t |}|dkr t d|dkr t d| |j }|dvr t d ||j||||||d |j|<|S) aAdd a seasonal component with specified period, number of Fourier components, and prior scale. Increasing the number of Fourier components allows the seasonality to change more quickly (at risk of overfitting). Default values for yearly and weekly seasonalities are 10 and 3 respectively. Increasing prior scale will allow this seasonality component more flexibility, decreasing will dampen it. If not provided, will use the seasonality_prior_scale provided on Prophet initialization (defaults to 10). Mode can be specified as either 'additive' or 'multiplicative'. If not specified, self.seasonality_mode will be used (defaults to additive). Additive means the seasonality will be added to the trend, multiplicative means it will multiply the trend. If condition_name is provided, the dataframe passed to `fit` and `predict` should have a column with the specified condition_name containing booleans which decides when to apply seasonality. Parameters ---------- name: string name of the seasonality component. period: float number of days in one period. fourier_order: int number of Fourier components to use. prior_scale: optional float prior scale for this component. mode: optional 'additive' or 'multiplicative' condition_name: string name of the seasonality condition. Returns ------- The prophet object. z1Seasonality must be added prior to model fitting.)rtrurvF)rrrzFourier Order must be > 0raz+mode must be "additive" or "multiplicative"r fourier_orderrr r)rBrRrlr0r/r6r-r=)rJrrrrr rrs rLadd_seasonalityzProphet.add_seasonalitysH << #CE E 4 4  % %d % F  --B{#B 767 7 A 89 9 <((D 5 5JK K  %  % %n 5*, $ 4  rNc|j tdt|D]}|j|d|j0t j dj|j|||_|S)aPAdd in built-in holidays for the specified country. These holidays will be included in addition to any specified on model initialization. Holidays will be calculated for arbitrary date ranges in the history and future. See the online documentation for the list of countries with built-in holidays. Built-in country holidays can only be set for a single country. Parameters ---------- country_name: Name of the country, like 'UnitedStates' or 'US' Returns ------- The prophet object. z6Country holidays must be added prior to model fitting.Fr_zHChanging country holidays from {country_holidays!r} to {country_name!r}.)r? country_name)rBrRrrlr?rPwarningr)rJrrs rLadd_country_holidayszProphet.add_country_holidayss( << #H &l3 BD  % %d5 % A B  , NN$%)%:%:!- !- rNc Zg}g}ggd}|jjD]\}}|j|d|d|d|}|d d|||d<|j||j |dg|j dz||d j||j |d}t|dkDrY|j|d|\}} } |j||j | ||jj | |jjD]W\}}|jtj|||j|d||d j|Yt|dk(rW|jtjd tj|j di|jd tj|d }|j!||\} }||| |fS) aDataframe with seasonality features. Includes seasonality features, holiday features, and added regressors. Parameters ---------- df: pd.DataFrame with dates for computing seasonality features and any added regressors. Returns ------- pd.DataFrame with regression features. list of prior scales for each column of the features dataframe. Dataframe with indicators for which regression components correspond to which columns. Dictionary with keys 'additive' and 'multiplicative' listing the component names for each mode of seasonality. rarrrrrrr[r rwraxis)r=rrappendrrrr%rr.r>r"rfrrwrregressor_column_matrix) rJrseasonal_featuresrmodesrrrr,holiday_priorsrcomponent_colss rLmake_all_seasonality_featuresz%Prophet.make_all_seasonality_featuressE& 26 --335 .KD%554ho& H %&29:"U#34556  $ $X .   }%&):: < %- ' ' - .33BtH= x=1 **2d8X> 4Hnm  $ $X .    / $$$ % , ,] ; 00668 .KD%  $ $R\\"T(%; <   m 4 5 %- ' ' - .  !Q &  $ $ grxx '<=> @    #II&7a@ $ < < u! !,EErNc <tjtj|jd|j Dcgc]}|j ddc}d}|j+|j|d|jj}dD]}|j||dz||}|jjDcgc]\}}|d|k(r|}}}|j|d |z|}||j|dz||jd |z||jjdtj|d |d jd } d D] } | | vsd| | <| j!ddddt#| d| dzdkDr t%d|j&?| |j&j } | j)|j&s t%d| |fScc}wcc}}w)a#Dataframe indicating which columns of the feature matrix correspond to which seasonality/regressor components. Includes combination components, like 'additive_terms'. These combination components will be added to the 'modes' input. Parameters ---------- seasonal_features: Constructed seasonal features dataframe modes: Dictionary with keys 'additive' and 'multiplicative' listing the component names for each mode of seasonality. Returns ------- component_cols: A binary indicator dataframe with columns seasonal components and rows columns in seasonal_features. Entry is 1 if that columns is used in that component. modes: Updated input with combination components. r[rqrcol componentr,ra_termsr extra_regressors_r r!)level)rsr{rwTignore)rrerrorsrsr{z&A bug occurred in seasonal components.z*A bug occurred in constructing regressors.)r"rfrarangerrrrFadd_group_componentrkr>rrr.crosstab sort_indexrrirRrDequals) rJrrr componentsr rrregressors_by_moderrs rLrzProphet.regressor_column_matrixSsX(\\99.44Q78/@/H/H*+ "1%#    # # /11J(@(@(G(G(IKJ3 ;D11D8OU4[J#'"7"7"="="?"a=D(" "11/$68JLJ $K  th / $K  2T9 : ; d  !((4 u z+6 *5* ! ? )D>)'(t$ ) G!T(K /0345 689 :DE E  $ $ 0+D,E,E,M,MNN!(()B)BC LMMu$$W"s H Hc||djt|j}|dj}t |dkDr/t j ||d}t j||g}|S)a[Adds a component with given name that contains all of the components in group. Parameters ---------- components: Dataframe with components. name: Name of new group component. group: List of components that form the group. Returns ------- Dataframe with components. r!r rr)rrrrkr%r"rfr)rJr,rgroupnew_comp group_colss rLr(zProphet.add_group_componentsuj5::3u:FGLLNe_++- z?Q ||JT$JKHJ#9:JrNc|dk(rdd}||jvr'tjdj||S|r'tjdj||S|}|S|dur|}|S|durd}|St |}|S)aGet number of fourier components for built-in seasonalities. Parameters ---------- name: string name of the seasonality component. arg: 'auto', True, False, or number of fourier components as provided. auto_disable: bool if seasonality should be disabled when 'auto'. default_order: int default fourier order Returns ------- Number of fourier components, or 0 for disabled. rrzQFound custom seasonality named {name!r}, disabling built-in {name!r} seasonality.rzXDisabling {name} seasonality. Run prophet with {name}_seasonality=True to override this.TF)r=rPrrre)rJrarg auto_disable default_orderrs rLparse_seasonality_argszProphet.parse_seasonality_argss &=Mt))) 55;VV5F$ @VV&!.  D[)M  E\M HMrNc|jdj}|jdj}|jdj}|j|j j dj}||z tjdk}|jd|j|d}|dkDr)d||j|jdd |jd<||z tjd kxs|tjd k\}|jd |j|d}|dkDr)d||j|jdd |jd <||z tjd kxs|tjd k\}|jd|j|d}|dkDr*d ||j|jdd |jd<yy)aSet seasonalities that were left on auto. Turns on yearly seasonality if there is >=2 years of history. Turns on weekly seasonality if there is >=2 weeks of history, and the spacing between dates in the history is <7 days. Turns on daily seasonality if there is >=2 days of history, and the spacing between dates in the history is <1 day. rrirrv gv@Nr r)weeksr[rurt)rBrjridiffrrnonzeror" Timedeltar7r)r0r-r=r*r+) rJfirstlastrmin_dtyearly_disablerweekly_disable daily_disables rLset_auto_seasonalitieszProphet.set_auto_seasonalitiess T"&&(||D!%%' \\$  $ $ &**,Q/0446 #(>>33 d--~rC 1  !.#;;--"& ,D  x ( %<",,Q*??<!R\\%:: 33 d--~qB 1 !.#;;--"& ,D  x (,1)==: BLLa$88 33 T++]A? 1 !.#;;--"& +D  w ' rNc@|dj|dj}}|dj||dj|z }|dj||dj|z |z }|dj|||dj|zz }||fS)aInitialize linear growth. Provides a strong initialization for linear growth by calculating the growth and offset parameters that pass the function through the first and last points in the time series. Parameters ---------- df: pd.DataFrame with columns ds (date), y_scaled (scaled time series), and t (scaled time). Returns ------- A tuple (k, m) with the rate (k) and offset (m) of the linear growth function. rrr)idxminidxmaxr)ri0i1Tkms rLlinear_growth_initzProphet.linear_growth_inits$D"BtHOO$5B sGLL r#w||B/ /  ^  $r*~':':2'> >! C zN   #a"S',,r*:&: :1v rNc^|dj|dj}}|dj||dj|z }|dj|}|dj|}td|zt d|z|dj|}td|zt d|z|dj|}||z }||z } t || z dkrd|z}t j|dz } t j| dz } | |z| | z z } | | z |z } | | fS) a'Initialize logistic growth. Provides a strong initialization for logistic growth by calculating the growth and offset parameters that pass the function through the first and last points in the time series. Parameters ---------- df: pd.DataFrame with columns ds (date), cap_scaled (scaled capacity), y_scaled (scaled time series), and t (scaled time). Returns ------- A tuple (k, m) with the rate (k) and offset (m) of the logistic growth function. rrrg{Gz?gGz?rg?r[)rIrJrrirjrrlog)rrKrLrMC0C1y0y1r0r1L0L1rOrNs rLlogistic_growth_initzProphet.logistic_growth_inits>$D"BtHOO$5B sGLL r#w||B/ /  " "2 &   " "2 & Cr 2j>+>+>r+BC D Cr 2j>+>+>r+BC D "W "W rBw<4 B VVBF^ VVBF^ Fb2g  "WM1v rNc4d}|dj}||fS)aInitialize flat growth. Provides a strong initialization for flat growth. Sets the growth to 0 and offset parameter as mean of history y_scaled values. Parameters ---------- df: pd.DataFrame with columns ds (date), y_scaled (scaled time series), and t (scaled time). Returns ------- A tuple (k, m) with the rate (k) and offset (m) of the linear growth function. rr)r)rrNrOs rLflat_growth_initzProphet.flat_growth_initIs$"  zN   !!t rNrc d|vsd|vr td||djj}|jddkr tdt j t j |djdj|_ |j|d |_ |j|j|j\}}}}||_||_t!||_|j%|j&d vr-t)j*|jjd}n|jd }t-|jjdt/|j0|jd |j2t4|j&j7j8|jd |jd|j0|||d|d| S)a Reformats historical data, standardizes y and extra regressors, sets seasonalities and changepoints. Saves the preprocessed data to the instantiated object, and also returns the relevant components as a ModelInputData object. rr~zPDataframe must have columns "ds" and "y" with the dates and values respectively.rrz'Dataframe has less than 2 non-NaN rows.rT)r)rXrZrr[rrrsr{) rMSKtautrend_indicatorr~rt_changeXsigmass_as_mr)r6notnullrrr"r$r#rkrrCrrBrGrrDrErrGrr rrwrr%r<r1ruppervalue) rJrkwargsrBrrrrrs rL preprocesszProphet.preprocess^s N2 ' RW__&',,. == a FG G^^BIIbhoo6Gd,ST``b++Gt+L  ##%  . .t|| < ?<$2!$"6*  ;;, ,((4<<--a01C,,|,Cll  #$%%&%%a(,,*4;;+<+<+>?EEll:&ll3((/056  rNnum_total_regressorsc|jdk(r|j|j\}}n[|jdk(r|j|j\}}n-|jdk(r|j |j\}}t t j|jt j|dS)z Calculates initial parameters for the model based on the preprocessed history. Parameters ---------- num_total_regressors: the count of seasonality fourier components plus holidays plus extra regressors. rXrZrYr)rNrOdeltabeta sigma_obs) r rPrBr]r[rr zeros_liker<rw)rJrmrNrOs rLcalculate_initial_paramsz Prophet.calculate_initial_paramss ;;( "**4<<8DAq [[F "((6DAq [[J &,,T\\:DAq-- 3 34./   rNc t|j td|j|fi|}|j|j}t j |}t j |}|jdj|jdjk(rv|jdk(s|jdk(rX||_ d|jd<|jD]2}tj|j|g|j|<4na|jdkDr/|jj|||jfi||_ n#|jj ||fi||_ |jj"|_t%|j&dk(r|jd|jd j)d z|jd<tj*|jd j,j)d |jd <|S) aFit the Prophet model. This sets self.params to contain the fitted model parameters. It is a dictionary parameter names as keys and the following items: k (Mx1 array): M posterior samples of the initial slope. m (Mx1 array): The initial intercept. delta (MxN array): The slope change at each of N changepoints. beta (MxK matrix): Coefficients for K seasonality features. sigma_obs (Mx1 array): Noise level. Note that M=1 if MAP estimation. Parameters ---------- df: pd.DataFrame containing the history. Must have columns ds (date type) and y, the time series. If self.growth is 'logistic', then df must also have a column cap that specifies the capacity at each ds. kwargs: Additional arguments passed to the optimizing or sampling functions in Stan. Returns ------- The fitted Prophet object. z>Prophet object can only be fit once. Instantiate a new object.r~rXrZg& .>rqrrNror)rr[)rBrRrlrsr` dataclassesasdictrjrir rArrr3rKsamplingfitr@r%r!reshaperwr)rJrrk model_inputsinitial_paramsdat stan_initpars rLrxz Prophet.fits2 << #89 9'tr4V4 66|~~F  .&&~6 <<  "dll3&7&;&;&= =(DKK6,A#DK'+DKK ${{ @#%88T[[-=,>#? C  @    "4$++44YTEVEVaZ`aDK/$++// 3I&IDK))22 t  !Q & C 4;;w#7#?#?#CC KK %'HHT[[-A-G-G$H'.ww'7 KK  rN vectorizedcF|j td||jj}n<|jddk(r t d|j |j}|j ||d<|j|}|jr|j||}nd}ddg}d|vr|jd|jr|jdtj||||fd }|dd |d zz|d z|d <|S)aUPredict using the prophet model. Parameters ---------- df: pd.DataFrame with dates for predictions (column ds), and capacity (column cap) if logistic growth. If not provided, predictions are made on the history. vectorized: Whether to use a vectorized method to compute uncertainty intervals. Suggest using True (the default) for much faster runtimes in most cases, except when (growth = 'logistic' and mcmc_samples > 0). Returns ------- A pd.DataFrame with the forecast components. NModel has not been fit.rzDataframe has no rows.rrrrrr[rr{rsry)rBrRrrr6r predict_trendpredict_seasonal_componentsr5predict_uncertaintyrr:r"r)rJrrseasonal_components intervalscolsdf2s rLpredictzProphet.predicts, << 56 6 :""$Bxx{a !9::%%bggi0B((,7 ">>rB  # #00Z@IIg B; KK     KK iiD9.ABKG C(>$? ?@&'( F  rNc|dddf|dk|z}|jd|z}|| zjd|z}||z|zS)aEvaluate the piecewise linear function. Parameters ---------- t: np.array of times on which the function is evaluated. deltas: np.array of rate changes at each changepoint. k: Float initial rate. m: Float initial offset. changepoint_ts: np.array of changepoint times. Returns ------- Vector y(t). N).Nr[r)sum)rdeltasrNrOchangepoint_tsdeltas_tk_tm_ts rLpiecewise_linearzProphet.piecewise_linear sb #47+q|;vElll"Q&>/)..A.6:Qw}rNcXtjtj|tj||zf}tjt |}t |D]5\}} | |z tj|z d||||dzz z z||<7|tj|z} |tj|z} t |D]*\} } || k\} | | xx|| z cc<| | xx|| z cc<,|dtj| || z zzz S)aEvaluate the piecewise logistic function. Parameters ---------- t: np.array of times on which the function is evaluated. cap: np.array of capacities at each t. deltas: np.array of rate changes at each changepoint. k: Float initial rate. m: Float initial offset. changepoint_ts: np.array of changepoint times. Returns ------- Vector y(t). r[) r concatenate atleast_1dcumsumrwr% enumerater ones_likeexp)rrrrNrOrk_cumgammasrUt_srrsindxs rLpiecewise_logisticzProphet.piecewise_logistic"s%$ a 0"))F2Ca2GHI#n-./ FAs1Wrvvf~-58eAEl224 1I  ",,q/!",,q/!/ #FAs8D I "I I "I #a"&&#S!12233rNc6|tj|z}|S)zEvaluate the flat trend function. Parameters ---------- t: np.array of times on which the function is evaluated. m: Float initial offset. Returns ------- Vector y(t). )rr)rrOrs rL flat_trendzProphet.flat_trendDs",,q/! rNc6tj|jd}tj|jd}tj|jdd}tj|d}|jdk(r |j |||||j }nV|jdk(r&|d }|j||||||j }n!|jd k(r|j||}|jz|d zS) zPredict trend using the prophet model. Parameters ---------- df: Prediction dataframe. Returns ------- Vector with trend on prediction dates. rNrOrorrrrXrYrrZr) rnanmeanrArr rr<rrr9)rJrrNrOrrrrrs rLrzProphet.predict_trendTs JJt{{3' ( JJt{{3' (DKK0q9 HHRW  ;;( "))!VQ4;N;NOE [[J &\"C++31d&9&9;E [[F "OOAq)Et||#bk11rNc~|j|\}}}}|jr*dd|jz zdz }dd|jzzdz }|j}i}|jD]} |j d|| jz} t j|| j} | |jdvr| |jz} t j| d|| <|js|j| d|| dz<|j| d|| d z<tj|S) zPredict seasonality components, holidays, and added regressors. Parameters ---------- df: Prediction dataframe. Returns ------- Dataframe with seasonal components. drrrprbr[rr|r})rr5r4rrrArmatmul transposerEr9r percentiler"rf) rJrr_rlower_pupper_prddatar!beta_ccomps rLrz#Prophet.predict_seasonal_componentspsT  . .r 2 01na  # #S4#6#667!;GS4#6#667!;G  $ $'// I[[(>)+D+K+KKF99Q 0 0 23DD00<< $ jjA6DO''-1__'.=.Y)*.2__'.=.Y)* ||D!!rNcZ|j||}dd|jz zdz }dd|jzzdz }i}dD]T}|j|||d|dj|<|j|||d|dj|<Vt j |S) aPrediction intervals for yhat and trend. Parameters ---------- df: Prediction dataframe. vectorized: Whether to use a vectorized method for generating future draws. Returns ------- Dataframe with uncertainty intervals. rrrryrrr[rz{}_lowerz{}_upper)sample_posterior_predictiver4rrr"rf)rJrr sim_valuesrrseriesrs rLrzProphet.predict_uncertaintys55b*E t2223a7t2223a7$ 2C-1__3q.=.2F:$$S) *-1__3q.=.2F:$$S) * 2 ||F##rNc h|jdjd}tdtt j |j t|z }|j|\}}}}ggd}t|D]y} |r|j||| |d|d|} n2t|Dcgc]}|j||| |d|d } }|D] } | D]} || j| | "{|jD]\} }t j||| <|Scc}w) aYProphet posterior predictive samples. Parameters ---------- df: Prediction dataframe. vectorized: Whether to use a vectorized method to generate future draws. Returns ------- Dictionary with posterior predictive samples for the forecast yhat and for the trend component. rNrr[rrsr{)rr iterationrfrg n_samples)rrrrfrg)rArrirerceilr5r/rrsample_model_vectorized sample_modelrr column_stack)rJrr n_iterations samp_per_iterrrrrrUsimsrsimrNvs rLrz#Prophet.sample_posterior_predictives{{3'--a0 As277  $ $u\': :$    . .r 2 01na!2. |$ 5A33&7&'78&'=>+ 4 %]3 %%*;"#*+;<*+AB &" 55CsO**3s845 5) 5.$$& /DAqOOA.JqM /s1#D/c|j||}|jd|}tj|j||jz|j z}tj|j||jz} |jd|} tj jd| |jd|j z} |d| zz|z| z|dS)aSimulate observations from the extrapolated generative model. Parameters ---------- df: Prediction dataframe. seasonal_features: pd.DataFrame of seasonal features. iteration: Int sampling iteration to use parameters from. s_a: Indicator vector for additive components s_m: Indicator vector for multiplicative components Returns ------- Dictionary with `yhat` and `trend`, each like df['t']. rprqrr[r) sample_predictive_trendrArrrr9randomnormalr) rJrrrrfrgrrrpXb_aXb_msigmanoises rLrzProphet.sample_models,,R;{{6"9-yy*11 *,.2ll;yy*114#**3DE K(3   E288A;7$,,FQX&-5  rNrrrfrgrc|jd|}tj|j||jz|jz}tj|j||jz} |j |||} |jd|} tj jd| | j|jz} g} t| | D]%\}}| j|d| zz|z|z|d'| S)zSimulate observations from the extrapolated generative model. Vectorized version of sample_model(). Returns ------- List (length n_samples) of dictionaries with arrays for trend and yhat, each ordered like df['t']. rprqrr[r) rArrrr9"sample_predictive_trend_vectorizedrrrzipr)rJrrrrfrgrrprrtrendsr noise_terms simulationsrrrs rLrzProphet.sample_model_vectorizeds {{6"9-yy*11szz)+-1\\:yy*114#**3DE88Y R K(3ii&&q%>M   4 LE5   T*T1E9    rNc|jd|}|jd|}|jd|}tj|d}|j}|dkDr;t |j }tj j||dz z} nd} | dkDr9dtj j| |dz zz} | jng} tjtj|dz} tj jd| | } tj|j | f} tj|| f}|jdk(r|j||||| }nL|jd k(r|d }|j!|||||| }n!|jd k(r|j#||}|j$z|d zS) a"Simulate the trend using the extrapolated generative model. Parameters ---------- df: Prediction dataframe. iteration: Int sampling iteration to use parameters from. Returns ------- np.array of simulated trend over df['t']. rNrOrorr[r:0yE>rXrYrrZr)rArrrir%r<rpoissonrandrrrlaplacerr rrrr9)rJrrrNrOrrrMr_ n_changeschangepoint_ts_newlambda_ deltas_newrrrrs rLrzProphet.sample_predictive_trends KK Y ' KK Y 'W%i0 HHRW  EEG q5D''(A ))!q1u+6II q=!"RYY^^I%>!a%%H!H   # # %!# ''"&&.)D0YY&&q'9= )<)<);)=> 45 ;;( "))!VQ>JE [[J &\"C++AsFAq,:rrBr<rArr _make_trend_shift_matrixr_logistic_uncertaintyrrrjrr)rJrrr uncertainties future_dfrrchange_likelihoodrrOrN mean_deltarpast_uncertaintys rLrzProphet._sample_uncertaintyms9 c7;;=A HHiR%9:MNKr#w{+I9~H!| ggin5::<  ggdll3&78==? #D$7$7 8; F [[))4F C +A C +A047J{{h&33J@QS[gp3q #  2 9 9q 9 A ,  *33J@QS[gp3q $ : :!!,/66$S>00% +!; ! & ")X)> ? ))#w{{}!#%88Yr#w!|8L,M#N "0@-/PWX Y rNr likelihood future_lengthc>tjj||f|k}tjjd||j}||z}tj tj t|df|gddddf}||zdz }|S)a! Creates a matrix of random trend shifts based on historical likelihood and size of shifts. Can be used for either linear or logistic trend shifts. Each row represents a different sample of a possible future, and each column is a time step into the future. )sizerr[Nrr)rruniformrrhstackrwr%)rrrrbool_slope_change shift_valuesrn_mats rLrz Prophet._make_trend_shift_matrixsII--I}3M-NQ[[yy((J=N=T=T(U .. 288SXqM2C89!SbS&As{a rNc|#tj|j}tjdd|z|}g}|D]/}|j tj ||kDdd1tj t|}|||<tj|jdd|d}||fS)z Creates a matrix of slope-deltas where these changes occured in training data according to the trained prophet obj rr[rr) rr>rr'rwhererwr%repeatry) rr<rn_rowr prev_timeidxs changepoint prev_deltass rL_make_historical_mat_timez!Prophet._make_historical_mat_times  ''&/..0KIIa[+> ) AK KK[!89! 1 represents the future) n_length: For each path, the number of future steps to simulate single_diff: The difference between each t step in the model context. Default None, inferred from t_time. Returns ------- A numpy array with shape (n_samples, n_length), representing the width of the uncertainty interval (standardized, not on the same scale as the actual data values) around 0. c|dk(}tj|tj|jdd}tjj |d||tj|jddddf|fS)Nrr[)rout)rrr'rmaximum accumulate)arrmaskidxs rLffillz,Prophet._logistic_uncertainty..ffillss!8D((D5"))DJJqM":A>C JJ ! !#A3 ! 7ryy1.q$w7<= =rNr[rrN)rr<r%rronesrrrrrrrrr)rJrrrNrOrrrrrhistorical_mathistorical_time full_t_timerk_cum_brrUrksrr sample_trendss rLrzProphet._logistic_uncertaintys@ >+/*H*HQUQdQdflnqrunvyD+E'nnnc2;nnov%>? 1q(9 :Q >biiX[bcNdghNhjk@lmtuv,s#syy|$ "AA"RVVF1bqb5M%BBAWQT]WQAX%666Br6F1a4L " zzqz!A%q8)*}5}}!}$q(!hYZ-8q2663$&3,*?#@@A }11q1999rNcd|j|j}|j||S)aSample from the posterior predictive distribution. Returns samples for the main estimate yhat, and for the trend component. The shape of each output will be (nforecast x nsamples), where nforecast is the number of points being forecasted (the number of rows in the input dataframe) and nsamples is the number of posterior samples drawn. This is the argument `uncertainty_samples` in the Prophet constructor, which defaults to 1000. Parameters ---------- df: Dataframe with dates for predictions (column ds), and capacity (column cap) if logistic growth. vectorized: Whether to use a vectorized method to compute possible draws. Suggest using True (the default) for much faster runtimes in most cases, except when (growth = 'logistic' and mcmc_samples > 0). Returns ------- Dictionary with keys "trend" and "yhat" containing posterior predictive samples for that component. )rrr)rJrrs rLpredictive_sampleszProphet.predictive_sampless-, ! !"'') ,//J??rNctj|jrtjntj}||g|i|S)aj We rely on np.nanpercentile in the rare instances where there are a small number of bad samples with MCMC that contain NaNs. However, since np.nanpercentile is far slower than np.percentile, we only fall back to it if the array contains NaNs. See https://github.com/facebook/prophet/issues/1310 for more details. )rrrh nanpercentiler)rJaargsrkfns rLrzProphet.percentiles;"$!!2R   !%d%f%%rNc|j td|;tj|jj d}| td|jj }tj ||dz|}|||kD}|d|}|r4tjtj|j|f}tjd|iS)aSimulate the trend using the extrapolated generative model. Parameters ---------- periods: Int number of periods to forecast forward. freq: Any valid frequency for pd.date_range, such as 'D' or 'M'. include_history: Boolean to include the historical dates in the data frame for predictions. Returns ------- pd.Dataframe that extends forward from the end of self.history for the requested number of periods. NrzUnable to infer `freq`r[)r7periodsfreqr) rCrRr" infer_freqrri date_rangerrrrf)rJrrinclude_history last_daters rLmake_future_dataframezProphet.make_future_dataframe&s    %56 6 <==!3!3!8!8!;BFK 4@D@D rNr)) __future__rrrrulogging collectionsrrrrdatetimer typingr r r numpyrpandasr" numpy.typingr prophet.make_holidaysrrprophet.modelsrrrr prophet.plotrr getLoggerrPsetLevelINFOrobjectrr2rNrLrBsnA@0$$ EWW0   9 % +e fe rN