[R] Transposing forecasts results from nnetar function and turn them into a data frame

[Rui Barradas]

Hello,

Is this what you want?


fun <- function(x, a){
	y <- as.numeric(as.character(x))
	z <- paste(a, names(x))
	data.frame(z, y)
}

dat2 <- as.data.frame(TSmodelForecast)

tmp <- lapply(seq_along(dat2), function(i) fun(dat2[[i]], names(dat2)[i]))
result <- do.call(rbind, tmp)

Hope this helps,

Rui Barradas

Em 17-03-2017 21:45, Rui Barradas escreveu:
> Hello,
>
> By running the code of the OP, I've come to the conclusion that
> print.forecast calls print.data.frame:
>
> forecast:::print.forecast
> function (x, ...)
> {
>      print(as.data.frame(x))
> }
> <bytecode: 0x0000000007ac9cd0>
> <environment: namespace:forecast>
>
> But I'm unable to get what the OP wants by call as.data.frame(...).
> Here is the code with some made up data.
>
> library(forecast)
> library(tseries)
>
> DAT <- data.frame(x = sample(100, 100, TRUE))
>
> TSdata <- ts(DAT[,1], start=c(1994,10), frequency=12)
> TSmodel <- nnetar(TSdata)
> TSmodelForecast <- forecast(TSmodel, h=24)
> TSmodelForecast
>
> str(as.data.frame(TSmodelForecast))
>
> Hope this helps,
>
> Rui Barradas
>
> Em 17-03-2017 20:48, Jim Lemon escreveu:
>> Hi Paul,
>> When manipulating any R object, the first thing to ascertain is what
>> it is:
>>
>> class(TSmodelForecast)
>>
>> should give you useful information.
>>
>> str(TSmodelForecast)
>>
>> should give you more. Because of the wealth of defined data structures
>> in R, it is difficult to manipulate them without this information. I
>> suspect that your output is something like a time series object, and
>> once that is known, it should not be too hard to display its contents
>> in the way you want.
>>
>> Jim
>>
>>
>> On Sat, Mar 18, 2017 at 7:12 AM, Paul Bernal <paulbernal07 at gmail.com>
>> wrote:
>>> Dear Jim,
>>>
>>> Hope you are doing great. I tried to do what you suggested but R send an
>>> error message saying that $ operator is invalid for atomic vectors.
>>>
>>> The format of the forecasts are as follows: forecasted years are as
>>> rows,
>>> and forecasted months are in columns what I want to do is to have two
>>> colums, one with the forecasted dates in (MMM-YYYY format) and the
>>> second
>>> column with the actual forecast results.
>>>
>>> The output that is giving me hard time is the forecast output from
>>> nnetar
>>> model.
>>>
>>> Thanks for your valuable support,
>>>
>>> Best of regards,
>>>
>>> Paul
>>>
>>>
>>> 2017-03-16 18:23 GMT-05:00 Jim Lemon <drjimlemon at gmail.com>:
>>>>
>>>> Hi Paul,
>>>> It looks like the information that is printed is in:
>>>>
>>>> TSModelForecast$mean
>>>>
>>>> If str(TSModelForecast$mean) returns something like a list with two
>>>> components, you can probably use something like this:
>>>>
>>>> paste(format(TSModelForecast$mean$Date,"%b-%Y"),
>>>>   TSModelForecast$mean$Forecast,sep="-",collapse="\n")
>>>>
>>>> It also might be in TSModelForecast$fitted
>>>>
>>>> Jim
>>>>
>>>>
>>>> On Fri, Mar 17, 2017 at 5:34 AM, Paul Bernal <paulbernal07 at gmail.com>
>>>> wrote:
>>>>> Dear friends,
>>>>>
>>>>> I am currently using R version 3.3.3 (64-bit) and used the following
>>>>> code
>>>>> to generate forecasts:
>>>>>
>>>>>> library(forecast)
>>>>>>
>>>>>> library(tseries)
>>>>>
>>>>>      ‘tseries’ version: 0.10-35
>>>>>
>>>>>      ‘tseries’ is a package for time series analysis and computational
>>>>> finance.
>>>>>
>>>>>      See ‘library(help="tseries")’ for details.
>>>>>
>>>>>
>>>>>> DAT<-read.csv("TrainingData.csv")
>>>>>>
>>>>>> TSdata<-ts(DAT[,1], start=c(1994,10), frequency=12)
>>>>>>
>>>>>> TSmodel<-nnetar(TSdata)
>>>>>>
>>>>>> TSmodelForecast<-forecast(TSmodel, h=24)
>>>>>>
>>>>>> TSmodelForecast
>>>>>
>>>>> The problem is that the output comes in this fashion:
>>>>>
>>>>>                  Jan    Feb    Mar    Apr    May    Jun     Jul    Aug
>>>>> Sep   Oct
>>>>>   2017        10      20      15      40     9         8         21
>>>>> 21
>>>>>      19     18
>>>>>   2018        34      15       7        6      10      11
>>>>>
>>>>> The format I would like to have is the following:
>>>>>
>>>>> Date                 Forecast
>>>>> Jan-2017               10
>>>>> Feb-2017               20
>>>>> Mar-2017               15
>>>>> Apr-2017                40
>>>>> May-2017               9
>>>>> Jun-2017                8
>>>>> Jul-2017                 21
>>>>> Aug-2017               21
>>>>> Sep-2017               19
>>>>> etc                          etc
>>>>>
>>>>> Is there a way to make the results look like this?
>>>>>
>>>>> Attached is a dataset as a reference.
>>>>>
>>>>> Best regards,
>>>>>
>>>>> Paul
>>>>> ______________________________________________
>>>>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>>> PLEASE do read the posting guide
>>>>> http://www.R-project.org/posting-guide.html
>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>
>>>
>>
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.