search

JuliaDynamics / TransitionsInTimeseries.jl

Transition Indicators / Early Warning Signals / Regime Shifts / Change Point Detection
MIT License
18 stars 5 forks source link

pvalues counting for surrogate significance is broken #65

Closed Datseris closed 7 months ago

Datseris commented 8 months ago

Have a look here:

https://juliadynamics.github.io/TransitionsInTimeseries.jl/dev/examples/logistic/#Significance-via-random-Fourier-surrogates

this exacmple used to be perfectly fine. Now, every single time point gets a significance flag. I believe this is because of our re-worked p-value counting code. That code is also over-engineered. We should simplify it back to its original, most simple possible version, because the performance of this code does not matter at all: it has miniscule impact on overall runtime.

JanJereczek commented 7 months ago

I have spotted the error. It came from how we newly handle nothing as an indicator. I will PR the fixed code soon!