Page 1 of 1

RQA on line

Posted: Tue Sep 9, 2014 12:03
by miquel
Hello!

I use the CRP toolbox to compute the recurrent plots and it is an excellent tool. However something illogical happens. I have determined RR, DET, etc.., of a time series of 4000 data with the tool on line for the size of neighbourhood e=0.3, giving RR=0.03 and for e=3, giving RR=0.13. I have also used the downloaded CRP toolbox with MATLAB and using the same input parameters (m=1, delay=1) and w=[], Euclidean norm, lmin=2, vmin=2, the results are for e=0.03, RR=0.011 and for e=0.3, RR=0.166. According to the results seems to be a difference in the RR values of a magnitude of ten comparing both methods (on line and installed CRP toolbox).
And increasing e, the value of RR is rapidly saturated (RR = 1) at values of e = 4 very lowers than 100%.
May be due to the different parameters used in calculation on line? or there are something that I'm doing wrongly.