Test for directional accuracy
rdb = dchange(y,yb,prt)
* y = vector or ts of observable or benchmark serie
* yb = vector or ts of competiting or forecast serie
* prt = %f if the user does not want to print the results (optional)
* rdb = a results tlist with
- rchange('meth') = 'direct. change'
- rchange('y') = benchmark or observable serie s
- rchange('yc') = competiting serie or forecasted series
- rchange('namey') = name of benchmark or observable series
- rchange('nameyc') = name of competiting series or forecasted series
- rchange('P') = proportion of time the sign of the observable value is correctly forecasted
- rchange('Pstar') = concordance index
- rchange('pvalue')= pvalue of the statistic
load(GROCERDIR+'\data\BusinessSurvey.dat'); // performs first differences dpp = delts(pp); dpf =delts(fp); bounds() rd = dchange('dpp','lagts(3,dpf)'); // Example taken from function dchange(): tests in french business survey if changes in // "level of future own production" are a good predictor of that of "level of past own production". | ![]() | ![]() |