[英]How to fit a regression model with ARIMA errors on the seasonally adjusted component of a time series (in R)?
I want to do these two things (combined) with a time series T:我想用时间序列 T 做这两件事(结合):
In other words, I want to obtain forecasts using the seasonally adjusted component of T integrating an external predictor and "adding back" the seasonality.换句话说,我想使用 T 的季节性调整组件集成外部预测器并“加回”季节性来获得预测。
I can do these two operations separately, but I can't get them to work in combination我可以分别做这两个操作,但我不能让它们结合起来工作
Here is some toy examples:以下是一些玩具示例:
First, load libraries and data:首先,加载库和数据:
library(forecast)
library(tsibble)
library(tibble)
library(tidyverse)
library(fable)
library(feasts)
library(fabletools)
us_change <- readr::read_csv("https://otexts.com/fpp3/extrafiles/us_change.csv") %>%
mutate(Time = yearquarter(Time)) %>%
as_tsibble(index = Time)
Example of fit and forecast with seasonally adjusted component of T:带有 T 的季节性调整分量的拟合和预测示例:
model_def = decomposition_model(STL,
Consumption ~ season(window = 'periodic') + trend(window = 13),
ARIMA(season_adjust ~ PDQ(0,0,0)),
SNAIVE(season_year),
dcmp_args = list(robust=TRUE))
fit <- us_change %>% model(model_def)
report(fit)
forecast(fit, h=8) %>% autoplot(us_change)
Example of regression model with ARIMA errors (Income as predictor):带有 ARIMA 错误的回归 model 示例(收入作为预测变量):
model_def = ARIMA(Consumption ~ Income + PDQ(0,0,0))
fit <- us_change %>% model(model_def)
report(fit)
us_change_future <- new_data(us_change, 8) %>% mutate(Income = mean(us_change$Income))
forecast(fit, new_data = us_change_future) %>% autoplot(us_change)
These examples work, but I would like to do something like this:这些示例有效,但我想做这样的事情:
model_def = decomposition_model(STL,
Consumption ~ season(window = 'periodic') + trend(window = 13),
ARIMA(season_adjust ~ Income + PDQ(0,0,0)),
SNAIVE(season_year),
dcmp_args = list(robust=TRUE))
fit <- us_change %>% model(model_def)
report(fit)
us_change_future <- new_data(us_change, 8) %>% mutate(Income = mean(us_change$Income))
forecast(fit, new_data = us_change_future) %>% autoplot(us_change)
I get this output in the console:我在控制台中得到了这个 output:
> fit <- us_change %>% model(model_def)
Warning message:
1 error encountered for model_def
[1] object 'Income' not found
>
> report(fit)
Series: Consumption
Model: NULL model
NULL model>
So I tried doing this in decomposition_model:所以我尝试在分解模型中这样做:
model_def = decomposition_model(STL,
Consumption ~ season(window = 'periodic') + trend(window = 13),
ARIMA(season_adjust ~ us_change$Income + PDQ(0,0,0)),
SNAIVE(season_year),
dcmp_args = list(robust=TRUE))
No problem with the fit, but now I get an error in the forecast:合身没问题,但现在我在预测中遇到错误:
> forecast(fit, new_data = us_change_future) %>% autoplot(us_change)
Error in args_recycle(.l) : all(lengths == 1L | lengths == n) is not TRUE
In addition: Warning messages:
1: In cbind(xreg, intercept = intercept) :
number of rows of result is not a multiple of vector length (arg 2)
2: In z[[1L]] + xm :
longer object length is not a multiple of shorter object length
What am I doing wrong?我究竟做错了什么?
Nothing wrong with your code here, just something I hadn't considered people would do when making decomposition_model()
.您的代码在这里没有问题,只是我没有考虑过人们在制作时会做的事情decomposition_model()
。 I've updated the decomposition modelling method to include exogenous regressors so that they can be used in component models ( https://github.com/tidyverts/fabletools/commit/8dd505f6378327b8e93b8440ec17ecf9badf2561 ).我更新了分解建模方法以包含外生回归量,以便它们可以用于组件模型( https://github.com/tidyverts/fabletools/commit/8dd505f6378327b8e93b8440ec17ecf9badf2561 )。 If you update the package, your first attempt at modelling should work fine.如果您更新 package,您的第一次建模尝试应该可以正常工作。
As for why the second attempt didn't work, the forecast method is finding us_change$Income and using that as the exogenous regressor for the future forecasts.至于为什么第二次尝试没有成功,预测方法是找到 us_change$Income 并将其用作未来预测的外生回归量。 This value has the length of us_change
, which does not match the length of us_change_future
, leading to the (confusing) error.该值的长度为us_change
,与us_change_future
的长度不匹配,导致(混淆)错误。
Reprex:代表:
library(tidyverse)
library(tsibble)
library(fable)
library(feasts)
us_change <- readr::read_csv("https://otexts.com/fpp3/extrafiles/us_change.csv") %>%
mutate(Time = yearquarter(Time)) %>%
as_tsibble(index = Time)
model_def = decomposition_model(STL,
Consumption ~ season(window = 'periodic') + trend(window = 13),
ARIMA(season_adjust ~ Income + PDQ(0,0,0)),
SNAIVE(season_year),
dcmp_args = list(robust=TRUE))
fit <- us_change %>% model(model_def)
report(fit)
#> Series: Consumption
#> Model: STL decomposition model
#> Combination: season_adjust + season_year
#>
#> ========================================
#>
#> Series: season_adjust
#> Model: LM w/ ARIMA(1,0,2) errors
#>
#> Coefficients:
#> ar1 ma1 ma2 Income intercept
#> 0.6922 -0.5777 0.1975 0.2035 0.5993
#> s.e. 0.1163 0.1305 0.0755 0.0462 0.0883
#>
#> sigma^2 estimated as 0.3234: log likelihood=-157.39
#> AIC=326.77 AICc=327.24 BIC=346.16
#>
#> Series: season_year
#> Model: SNAIVE
#>
#> sigma^2: 0
us_change_future <- new_data(us_change, 8) %>% mutate(Income = mean(us_change$Income))
forecast(fit, new_data = us_change_future) %>% autoplot(us_change)
Created on 2019-10-09 by the reprex package (v0.2.1)由reprex package (v0.2.1) 于 2019 年 10 月 9 日创建
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.