differences between the weighted and non weighted models

[Tobi]

Hi Chris,

we struggle with the use of the ctmm package. We want to calculate the AKDE and tried it with the minireview example. The Problem we are facing is, that our calculated AKDE with weights = TRUE is the same as the AKDE with weights = FALSE.
Our data was collected over ~2 months with irregular sampling times from 5 sec to half a day.

So we wondered why aren’t there (in our case) any differences between the weighted and non weighted models? 

Greetings,
Tobi

[Christen Fleming]

Hi Tobi,

If the sampling is irregular in a kind of uniformly random way, but the effective sample size is very large, then the improvement to AKDE can be very minimal. On the other hand, if you had switched between 5-second sampling and 12-hour sampling at one month into the data collection, or if you had a shorter tracking duration compared to the animal's home-range crossing timescale, then the weights would still be impactful.

https://esajournals.onlinelibrary.wiley.com/doi/10.1002/eap.1704

We haven't yet examined the impact on RSF estimation, for cases such as yours.

Best,

Chris