I am trying to forecast multiple timeseries using auto.arima with the output of units in a dataframe.
I have this sample data set:
Name <- c(rep("Tom",24),
rep("Jake",12),
rep("Jill",12))
Product <- c(rep("Orange",12),rep("Apples",12),
rep("Orange",12),rep("Apples",12))
Quarter <- c(rep(c("2013-03-31","2013-06-30","2013-09-30","2013-12-31",
"2014-03-31","2014-06-30","2014-09-30","2014-12-31",
"2015-03-31","2015-06-30","2015-09-30","2015-12-31"),4))
Units <- c(1200,1500,1800,2600,1300,1500,1700,2600,1300,1700,1900,2700,300,400,800,900,
250,400,600,700,200,300,0,0,100,250,500,600,90,200,400,450,50,0,0,0,1200,0,0,0,
1500,1700,1800,2600,1600,1900,2000,3000)
data <- cbind(Name,Product,Quarter,Units)
data <- as.data.frame(data)
data$Name_Product <- paste(data$Name,data$Product)
data <- select(data,Quarter,Units,Name_Product)
data <- spread(data,Name_Product,Units)
data2 <- ts(data[,-1], frequency=4, start=2013)
I am trying to forecast the batch data above by Name and Product using the following script:
h <- 12
ns <- ncol(data)
arimaforecast <- matrix(NA, nrow=h, ncol=ns)
for (i in 1:ns){
arimaforecast[,i] <- forecast(auto.arima(data2[,i]), h=h,)$mean}
My problem is that I keep getting the error "Error in '[.default'(data2, , i) : subscript out of bounds" and am unable to create a forecast similar to using the forecast() function from Rob Hyndman's batch forecasting approach but using the auto.arima function
I'm not sure where to go from here or if I am using the correct method. Any help is appreciated, thanks!
Change this line of your code:
ns <- ncol(data)
to:
ns <- ncol(data2)
Your loop was cycling through the 5 columns in your original data.frame instead of the four from your new time series matrix.