How do I remove the season from a series? (seasonal adjustment with X-13, R code to take along)

In brief

How do I remove the season from a monthly or quarterly series? X-13 does not run in the browser, so Statable does not calculate the seasonal adjustment itself. It creates the series as a CSV file and an R script to take along, run in R with the package seasonal.

How it works in Statable

1. In the time series analysis open the section "Seasonal adjustment with X-13 (in R on your own computer)".
2. Choose the method: SEATS (default) or X-11.
3. "Save the series as CSV" and "Save the R script".
4. Put both files in one folder and run the script in R. It writes a second CSV with the adjusted series.

Not the same as the classical decomposition

The decomposition in the time series analysis (trend, season, remainder) is not a seasonal adjustment with X-13 and is not called that. It stays as it is.

When it is offered

• Only for monthly or quarterly values.
• The series must have no gaps. Statable fills nothing in.
• At least 3 full years are needed: 36 monthly values or 12 quarterly values.
• If any of this does not apply, a sentence with the reason appears instead of the buttons.

All steps in detail

1. Choose the method: SEATS (default, the standard of the package seasonal) or X-11.
2. "Save the series as CSV" and "Save the R script".
3. Put both files in the same folder and run the script in R. For this R needs the package seasonal (with x13binary).
4. The script writes a second CSV file with the adjusted series. It can be opened in Statable again.

The CSV file contains the point in time and the value, with a point as decimal sign, in UTF-8. With "View the R script" the script can be read before saving.

Warnings

• Statable does not calculate this. The result is produced in R on your own computer, and Statable has not checked it.
• Calendar: seasonal tests the standard trading-day and Easter effects. Austrian public holidays are not included; the ESS guidelines recommend national calendars. If you need them, add your own regressors via xreg.
• The model choice of X-13 is automatic. summary(m) and the diagnostics need to be read.

Typical examples

• Remove the yearly pattern from a monthly unemployment rate before the development is interpreted.
• Adjust quarterly sales over several years and compare them with the original series.

For specialists

library(seasonal)
d <- read.csv("reihe.csv")
y <- ts(d$wert, start = c(1949, 1), frequency = 12)   # 12 bei Monaten, 4 bei Quartalen
m <- seas(y)                                           # bei X-11: seas(y, x11 = "")
summary(m)
plot(m)
bereinigt <- final(m)

As a check, seas(AirPassengers) gives the model (0 1 1)(0 1 1); the first adjusted values are 122.713, 124.766 and 125.073.