search

MattWillFlood / EntropyHub

An open-source toolkit for entropic data analysis.
https://www.entropyhub.xyz
Apache License 2.0
146 stars 19 forks source link

BubbEn output interpretation #6

Closed ejc043 closed 1 year ago

ejc043 commented 1 year ago

Thank you for this wonderful package!! For the BubbEn function for some values of m, there are multiple outputs of the estimated bubble entropy Bubb. How can we interpret the multiple outputs? Would this be the average? Thank you!

MattWillFlood commented 1 year ago

Hi @ejc043 !

Thank you for your question : ) BubbEn() returns two variables:

  1. Bubb - the bubble entropy estimates The number of values returned in Bubb corresponds to the embedding dimension (m) used to estimate Bubble entropy. Bubb will always have m-1 values, so if m=2, Bubb will have 1 value; if m=5, Bubb will have 4 values. Therefore, each value in Bubb is the bubble entropy estimate calculated using each dimension from 2:m. If you use m>2, the last value in Bubb will be the value for that m.

    So if you call: Bubb, H = BubbEn(Signal, m= 7) Bubb[6] will be the bubble entropy estimate for m==7, Bubb[4] will be the bubble entropy estimate for m==5, Bubb[1] will be the bubble entropy estimate for m==2.

    In most cases, you just want the last value in Bubb in your analysis.

  2. H - the conditional Renyi entropy estimates Similarly, the number of values returned in H also corresponds to m. H will always have m+1 values, corresponding to the conditional Renyi entropy estimates for each value from 1:m+1. Note, the first value of H will always be zero.

For more info, see the paper by Manis et al.

Apologies for this confusion. I will try to update the EntropyHub Guide and the relative documentation to make this more clear.

I hope this helps! Thanks, Matt

ejc043 commented 1 year ago

Hi @MattWillFlood, Thank you for the fast response! Yes that makes sense. Thank you!! Eunice