SAS ETS. Лекция 3 (1185367), страница 2

Просмтор этого файла доступен только зарегистрированным пользователям. Но у нас супер быстрая регистрация: достаточно только электронной почты!

Текст из файла (страница 2)

A l l r i g h t s r es er v e d .PROC SEVERITY:PARAMETERINITIALIZATION•You can specify your own initial values•INIT= option in the DIST statement• INEST= option in the PROC SEVERITY statement• Programmatic initialization that uses the sample values of the response variable orthe sample EDF•Predefined distributions have a default initialization method•Method of moments for most• Method of percentile matching for Weibull• Approximate maximum likelihood method for gammaC op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .EXAMPLE:Suppose you want to fit a distribution model other than one of the predefinedones available to you. Suppose you want to define a model for the Gaussiandistribution with the following typical parameterization of the PDF and CDF:For PROC SEVERITY, a distribution model consists of a set of functions andsubroutines that are defined with the FCMP procedure.C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c .

A l l r i g h t s r es er v e d .МОДЕЛЬ НЕНАБЛЮДАЕМЫХ КОМПОНЕНТPROC UCMC op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .SAS/ETS МОДЕЛЬ НЕНАБЛЮДАЕМЫХ КОМПОНЕНТМодель ненаблюдаемых компонент можно рассматривать как модельмножественной регрессии с изменяющимися во временикоэффициентами. Она основана на следующих принципах:•Временной ряд может быть разложен на трендовую, сезонную ициклическую компоненты.•Модели временных рядов, которые дают равный вес ближним иотдаленным наблюдениям, неприменимы.C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .SAS/ETS МОДЕЛЬ НЕНАБЛЮДАЕМЫХ КОМПОНЕНТA UCM decomposes the response series into components such as trend,seasons, cycles, and the regression effects due to predictor series.

The followingmodel shows a possible scenario:andC op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .represent the trend, seasonal, and cyclical components, respectively.SAS/ETS MODELING THE TRENDThe UCM procedure offers two ways to model the trend component:• The random walk (RW) model (implies that the trend remains roughly constantthroughout the life of the series without any persistent upward or downward drift):•In the second model the trend is modeled as a locally linear time trend (LLT),consisting of both the level and slope:C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c .

A l l r i g h t s r es er v e d .SAS/ETS MODELING A CYCLEA deterministic cyclewith frequency,:If the argument t is measured on a continuous scale, thenis a periodic functionwith periodamplitudeand phaseEquivalently, the cycle can be written in terms of the amplitude and phase as:C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .SAS/ETS MODELING SEASONSThe seasonal fluctuations are a common source of variation in time series data.These fluctuations arise because of the regular changes in seasons or some otherperiodic events.

The seasonal effects are regarded as corrections to the generaltrend of the series due to the seasonal variations, and these effects sum to zerowhen summed over the full season cycle.C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .SAS/ETS PROC UCMPROC UCM <options> ;AUTOREG <options> ;BLOCKSEASON options ;BY variables ;The UCM procedure uses thefollowing statements:CYCLE <options> ;DEPLAG options ;ESTIMATE <options> ;FORECAST <options> ;ID variable options ;IRREGULAR <options> ;The PROC UCM and MODELstatements are required. In addition,the model must contain at least onecomponent with nonzerodisturbance variance.LEVEL <options> ;MODEL dependent variable <= regressors> ;NLOPTIONS options ;PERFORMANCE options ;OUTLIER options ;RANDOMREG regressors </ options> ;SEASON options ;SLOPE <options> ;SPLINEREG regressor <options> ;SPLINESEASON options ;C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c .

A l l r i g h t s r es er v e d .Спасибо за внимание!C op yr i g h t © 2 0 1 2 , S A S I n s t i t u t e I n c . A l l r i g h t s r es er v e d .SAS.com.

Характеристики

Список файлов лекций