CSBLatvia / SA-indirect-quality

Quality evaluation for indirect seasonal adjustment

Quality evaluation for indirect seasonal adjustment

Aim

To provide a package for testing and quality evaluation for the indirect seasonal adjustment according to the Handbook on Seasonal Adjustment (2018, Section 18.3.2 "Quality Measures").

Steps

1. Functions for quality criteria calculation for two time series – directly and indirectly adjusted (tsda and tsia). Could be the seasonally adjusted series, the trend-cycle estimates and the seasonal components.
   • Input: two sets of time series (two ts objects with three time series – sa, t, i).
   • Parameters:
     • Length of time series in years (could be decimal): full (NULL), defined time period (for example last 3 years).
   • Output:
     • named list of values for quality criteria.
     • plots
2. Functions for running tests on spreadsheets of time series (tssda and tssia).
   • Input: two tables with time series.
   • Output:
     • table (or list) with quality criteria.
     • plots
3. Functions for doing seasonal adjuestment (with RJDemetra) and aggregation of time series (with KIX) and computing quality criteria for all time series.
   • Input: time series, weights and aggregation specification (how to define in R? formulas?).
   • Output:
     • table (or list) with quality criteria.
     • plots

History

Initialy developed during the Hackathon on RJDemetra in Frankfurt - 1st and 2nd July 2019.