CODEWITHSUNDEEP
rforecastingforecast
Cauder
Cauder

Reputation: 2627

The length of the train and test sets are different from the length of the actual set

I'm using test_forecast and I have my code setup like this

df_train <- df[1:20] 
df_test <- df[21:nrow(df)] 

test_forecast(actual = df, 
              forecast.obj = forecast,
              train = df_train, 
              test = df_test)

Every row in the df is accounted for in df train and df test.

But, it gives me this error

Error in test_forecast(actual = df, forecast.obj = forecast,  : 
  The length of the train and test sets are different from the length of the actual set

df is a data table, but I also tried this converting all the objects to data frames and that didn't work

Upvotes: 0

Views: 500

Answers (1)

Waldi
Waldi

Reputation: 41260

I tried with a data frame and a data table and got exactly the same error message.

After reading carefully test_forecast documentation, I found :

actual: The full time series object (supports "ts", "zoo" and "xts" formats).

The conclusion is that test_forecast isn't supposed to work with data.frames / data.tables.

As you didn't provide the data you're using, I tried this example which worked:

library(TSstudio)
ts <- USgas
ts_par <- ts_split(ts, sample.out = 20)

train <- ts_par$train

test <- ts_par$test

ts_info(train)
ts_info(test)

library(forecast)

md <- tslm(train ~ season + trend)

fc <- forecast(md, h = 20)
test_forecast(actual = ts,
              forecast.obj = fc,
              test = test)

enter image description here

Another important point is that you shouldn't subset a timeseries as you would do with a data.table, because it becomes numeric, so that it won't work with test_forecast:

class(USgas[1:20])
[1] "numeric"

Upvotes: 1

Related Questions