Transfer entropy normalization

[Katja Sangati]

Hi,

I'm trying to compare agent-based simulation results, in which behaviors were evolved while maximizing different entropy metrics. In order to make the comparison, I need different metric values to be all within the same range (preferably 0 to 1).

Is there an option within the JIDT package to compute TE values normalized to the [0, 1] range?

I found this discussion on stackexchange but it's a bit beyond my skillset to evaluate and implement the proposed solutions.

Thank you in advance!

Best,

Katja

[Joseph Lizier]

Hi Katja,

The discussion you point to lays out the theoretical considerations quite well, particularly Greg Ver Steeg's well thought out comments. As they say, such normalisation really only makes sense if you are dealing with discrete variables rather than continuous ones. You then have a number of options to make the normalisation as they say.

In short, there is no option to do this in JIDT -- you would have to separately compute the quantity you want to normalise by, then divide your TE by that.

--joe
+61 408 186 901 (Au mobile)

--
JIDT homepage: https://github.com/jlizier/jidt
---
You received this message because you are subscribed to the Google Groups "Java Information Dynamics Toolkit (JIDT) discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jidt-discuss...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jidt-discuss/ae74dedc-ce08-48c5-82ac-ea13f8d216can%40googlegroups.com.