search

PengTao-HUST / crossmapy

implements several causal inference algorithms based on dynamical causality (DC) framework
MIT License
5 stars 0 forks source link

different embedding dimensions #2

Open jordanplanders opened 5 months ago

jordanplanders commented 5 months ago

Hi there,

I very much appreciated your paper and find the content of this repo very intriguing!

I can't see anywhere where you support different embedding dimensions for different variables (and relatedly, tau values). Is this because it was not necessary for your examples, or because it is methodologically insensible?

Thanks!

PengTao-HUST commented 4 months ago

Sorry for the late reply! Frankly speaking, although there are ready-made methods for tau values, they don't perform too well in practice, so the strategy I use is to try a few different tau values and then choose the one with the highest accuracy or the strongest interpretability out of them. I hope this can help you.