pucicu
commented
5 years ago I agree that keeping rqaentropy would be good.
Leonardi’s paper is meanwhile published.
names … good question. we could try to stay close to pyunicorn. but i don’t like the term “white_vert_entropy”, because that the RP points have colors white and black is not unique, we could also simply reverse the color and nothing should change from the mathematical point of view. therefore, instead of “white_vert_entropy” i think that “recurrence time entropy” (rectimeEntropy) would be better.
Am 02.01.2019 um 23:31 schrieb Helios De Rosario notifications@github.com:
Following up #25 https://github.com/JuliaDynamics/RecurrenceAnalysis.jl/pull/25, particularly the issue about the entropy measures derived from recurrence plots.
In summary, there are three entropy-related parameters, which are the Shannon entropies of (1) the "black" diagonal lines (the usual RQA entropy, which is now calculated by rqaentropy), (2) the black vertical lines (Leonardi's paper to be published), and (3) the white vertical lines (recurrence time entropy).
Besides, recurrence plots can be used to estimate the Kolmogorov entropy, but that is another story I think.
I don't think that eventually removing rqaentropy due to the wealth of entropy-related parameters is a good idea. It is one of the parameters that I have seen reported most frequently, after the recurrence rate and determinism, so its calculation should be in the same level. Yes, it can be calculated from the histogram with some other generic entropy function, but all RQA parameters can eventually be calculated from the histogram and some simple function (actually simpler than the entropy), and moreover there are options like the minimum length that should be taken into account.
pyunicorn has different entropy parameters corresponding to the three that Norbert mentioned, cf. http://www.pik-potsdam.de/~donges/pyunicorn/api/timeseries/recurrence_plot.html http://www.pik-potsdam.de/~donges/pyunicorn/api/timeseries/recurrence_plot.html : diag_entropy, vert_entropy, and white_vert_entropy. Those names are clear and meaningful, but not really the style of Julia functions.
@pucicu https://github.com/pucicu: any suggestion for the names?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/JuliaDynamics/RecurrenceAnalysis.jl/issues/30, or mute the thread https://github.com/notifications/unsubscribe-auth/AA7WkNUvZilquRmRENjEg-Idd-KoAEmXks5u_TNDgaJpZM4ZnUyu.
heliosdrm
Following up #25, particularly the issue about the entropy measures derived from recurrence plots.
In summary, there are three entropy-related parameters, which are the Shannon entropies of (1) the "black" diagonal lines (the usual RQA entropy, which is now calculated by
rqaentropy), (2) the black vertical lines (Leonardi's paper to be published), and (3) the white vertical lines (recurrence time entropy).Besides, recurrence plots can be used to estimate the Kolmogorov entropy, but that is another story I think.
I don't think that eventually removing
rqaentropydue to the wealth of entropy-related parameters is a good idea. It is one of the parameters that I have seen reported most frequently, after the recurrence rate and determinism, so its calculation should be in the same level. Yes, it can be calculated from the histogram with some other generic entropy function, but all RQA parameters can eventually be calculated from the histogram and some simple function (actually simpler than the entropy), and moreover there are options like the minimum length that should be taken into account.pyunicorn has different entropy parameters corresponding to the three that Norbert mentioned, cf. http://www.pik-potsdam.de/~donges/pyunicorn/api/timeseries/recurrence_plot.html :
diag_entropy,vert_entropy, andwhite_vert_entropy. Those names are clear and meaningful, but not really the style of Julia functions.@pucicu: any suggestion for the names?