ctmm.fit() updates for small amounts of calibration data

[Christen Fleming]

There are new features in the development branch of ctmm on GitHub that are relevant to people with small amounts of calibration data. These features are demonstrated in the updated location-error vignette: https://ctmm-initiative.github.io/ctmm/articles/error.html#model-fitting-1 The related UERE object format has also been changed, so old objects will not be properly recognized by new versions of the package. uere(DATA) <- NULL will reset uncalibrated data properly, however.

Previously, the location error parameters calculated by uere.fit() were held fixed when running ctmm.fit(), whereas now their uncertainties are carried through and updated by ctmm.fit(). With adequate amounts of calibration data, this doesn't really do anything, but with small amounts it can be beneficial.

ctmm.fit() can now also fit multiple UERE parameters, though this is still highly inadvisable without calibration data.

Finally, in a somewhat contrived way, you can specify a prior for your UERE parameters, which is demonstrated in the updated vignette, where I specify a prior with 95% credible intervals of 3.5–17 meters for 3D GPS fixes at HDOP=1. This may or may not be appropriate for your device and I would first suggest taking a look the device appendix of https://www.biorxiv.org/content/10.1101/2020.06.12.130195v1 before doing that.

At some point these methods will be improved to be able to pool heterogeneous device calibration data into a population model and to model heavy tailed errors.

Best,

Chris