Open kohleth opened 5 years ago
@robjhyndman - is this repo/package still being maintained?
Are the features computed here equivalent to those in FEASTS - or are the two slowly diverging to make way for multi-series computations?
tsfeatures
is still being maintained, at least fixing bugs, etc., and we will continue to ensure it stays on CRAN. The features in feasts
may diverge as that is under more active development.
Hi,
I am trying to understand the logic behind
sampenc
. I figured it is a modification of douglas lake's code? But there seems to be 2 issues:p <- A[2]/B[1]
I think should bep<-A[M]/B[M-1]
issue 2 is, I don't quite get lake's code (is it correct?). I tried to follow his argument on https://www.physionet.org/content/sampen/1.0.0/ especially
But I couldn't verify this behaviour using some simple simulated series.
So I
pracma::sample_entropy
and did a comparison
So
pracma:;sample_entropy
(and my translation of the original paper) more or less agree while strongly disagreeing with lake's code (and thustsfeatures::sampenc
).But again, I couldn't follow lake's algorithm at all.