Seasonal Adjustment in R or Python

Question

Does anybody know of a routine to do seasonal adjustment in Python or even better, in R? Here is an example data-set (South African CPI), which tends to have spikes in the first few months of the year:

I would like to find the underlying pressures stripping out the seasonal factors, but I'd ideally like to use something fairly straightforward, built into either language, rather than interfacing or using outright an external package such as Demetra.

Answer 1

Step 1. Define the data.

(Obtained from http://www.statssa.gov.za/publications/P0141/P0141February2011.pdf)

CPI <- c(102.3, 103.1, 104.3, 105.7, 106.2, 106.6, 107, 108.2, 108.5, 108.9, 
        108.9, 108.9, 109.2, 109.5, 110.2, 111.1, 111.3, 111.5, 111.5, 
        112.2, 112.3, 112.4, 112.6, 112.8, 113, 113.5, 114.3)

Step 2. Calculate monthly change in index, and convert to time series object.

dCPI <- ts(CPI[-1] - CPI[-length(CPI)], start=2008, frequency=12)

Step 3. Use the function stl() to calculate seasonal, trend and residuals:

plot(stl(dCPI, "periodic"))