Difference between isotropic and anisotropic

[Marinka van Puijenbroek]

Hey Chris,  

So I have been working with your package and the results are really nice. We have gps tracking data of multiple individuals of bats and we have been able to create home ranges for each individual. 

For each of these individuals, we always selected the best model, but sometimes an isotropic or an anisotropic model was chosen. It is for me not really clear what the difference is between an isotropic and an anisotropic model. 

I read in the help file that isotropic or anisotropic determines whether the animal's covariance is circular or elliptical. For which data is isotropic the best method and when is anisotropic the best method. Furthermore, how does these two different methods affect the homerange, does it differ a lot between these two methods? 

Thanks very much for your help, 

Marinka 

[Christen Fleming]

Hi Marinka,

More elongated distributions will require anisotropic models to properly account for that in the KDE. Misapplying an isotropic model in this case would result in a biased estimate (too short along the long axis and too long along the short axis).

Relatively symmetric distributions will not support these extra parameters, and an isotropic model will be selected. Misapplying an anisotropic model in that case would result in extra statistical error from the unsupported parameters.

Best,

Chris

[Marinka van Puijenbroek]

Hey Chris, 

Thanks a lot for this clear answer. 

Cheers, 

Marinka 

Op woensdag 19 mei 2021 om 16:02:06 UTC+2 schreef chris.h...@gmail.com: