Figure 8. ACC data visualization including behaviour
classification results using a dynamic graph. White stork ACC data is
shown from segment 55 to 80 (cf Fig. 2b). Vertical black, dashed lines
separate different observed behaviour types, while vertical grey dotted
lines identify segments that have been incorrectly classified.
Conclusions
As demonstrated, the rabc package aims to assist researchers in
developing good animal behaviour classification models in an interactive
fashion. As human brains are extremely good at recognizing patterns, the
visualization of data and results can greatly assist the workflow
towards developing a behaviour classification model. Raw ACC data
visualization assists in the detection of aberrant associated behaviour
scores. Feature visualization helps researchers to understand how
different features distribute across behaviours and whether the current
behaviour set potentially needs adjustments, either by combining or by
splitting behaviours up. Finally, classification-result visualization
assists the understanding of misclassification patterns. Other than the
visualization functionalities, this package provides complete functions
to perform behaviour classification through XGboost, including: feature
calculation, feature selection, model hyperparameter tuning, model
training and validation and an output classifier for future ACC data
classification.
Given its unique aim and functionality the rabc package will be a
valuable addition to the growing array of R packages already available
for behaviour and movement analyses (Joo et al., 2020). It is worth
noting that the features calculated in this package can be further
extended if deemed neccessary. Users can develop additional features and
include these in the here described analyses and the ultimate generation
of a behaviour classification model. Although we only use XGboost as the
supervised machine learning model in this package, users can potentially
use the output from the rabc package as input to the “caret” package.
This will allow for the use of other machine learning models in
generating behaviour classification models such as for example decision
tree, support vector machine and random forest.