arimax warning. Why?

[espher1987]

Hi, i am getting problems with the last updated on cran. My goal is to get an Arimax model, so this is a basic example of the warning i get.

in the past. in forecast 8.4 using the next example.

library(forecast)
set.seed(1)
x <- rbinom(length(AirPassengers),1,0.5)
model <- Arima(AirPassengers,order = c(1,1,0),c(1,1,0),xreg = x)
forecast(model,xreg=c(1,1,1,0,0))

i dont get any error everything looks great, not warning either.

but in forecast 8.5
the same code give me this:

Warning message:
In forecast.Arima(model, xreg = c(1, 1, 1, 0, 0)) :
xreg contains different column names from the xreg used in training. Please check that the regressors are in the same order.

i must say i am getting the forecast information. But why this warning? based in 8.5 documentation xreg can be "A numerical vector".

same happens even if i do:

library(forecast)
set.seed(1)
x <- rbinom(length(AirPassengers),1,0.5)
model <- Arima(AirPassengers,order = c(1,1,0),c(1,1,0),xreg = x)

x <- c(1,1,1,0,0)
forecast(model,xreg=x)

Thanks in advance.

[mitchelloharawild]

Hi, This issue is a warning and there should be no subsequent issues with the output you are receiving. In the latest update we made some changes to exogenous regressors for ARIMA models to make it less likely for user error. One common issue was the naming of regressors in the model. As part of this, exogenous regressors now should be a named matrix of regressors. The names of the matrix provided to the forecast function should match the one used to train the model (so that correct columns are used for the appropriate xreg terms).

…

On Tue., 29 Jan. 2019, 3:18 am espher1987 ***@***.*** wrote: Hi, i am getting problems with the last updated on cran. My goal is to get an Arimax model, so this is a basic example of the warning i get. in the past. in forecast 3.4 using the next example. library(forecast) set.seed(1) x <- rbinom(length(AirPassengers),1,0.5) model <- Arima(AirPassengers,order = c(1,1,0),c(1,1,0),xreg = x) forecast(model,xreg=c(1,1,1,0,0)) i dont get any error everything looks great, not warning either. but in forecast 3.5 the same code give me this: Warning message: In forecast.Arima(model, xreg = c(1, 1, 1, 0, 0)) : xreg contains different column names from the xreg used in training. Please check that the regressors are in the same order. i must say i am getting the forecast information. But why this warning? same happens even if i do: library(forecast) set.seed(1) x <- rbinom(length(AirPassengers),1,0.5) model <- Arima(AirPassengers,order = c(1,1,0),c(1,1,0),xreg = x) x <- c(1,1,1,0,0) forecast(model,xreg=x) Thanks in advance. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#771>, or mute the thread <https://github.com/notifications/unsubscribe-auth/APYUl0pbOxapPiPCLeRxkktlQb10CDfqks5vH0zngaJpZM4aWl5z> .

[espher1987]

Thanks, but why this warning is showed even when you are not working with matrix. I use a numeric vector, so, a warning that tell me about "column names" and "regressors in same order" is weird. I get the idea if i use matrix, now i get it, but if i use a numeric vector, should this warning still be showed? Documentation says that numeric vector is allowed. I just feel a bit confuse. Thanks in advance.