Table 2 Methods for assigning accelerometer waveforms to behavioral categories based on waveform statistics
From: Observing the unwatchable through acceleration logging of animal behavior
| Method | Representative source(s) | |
|---|---|---|
| Manual | ||
| Reference patterning | [34, 98, 108, 127, 171] | |
| Fixed-threshold | [31, 57, 83, 161] | |
| Unsupervised machine learning algorithms | ||
| Cluster analysis | [164] | |
| Supervised machine learning algorithms | ||
| Classification and regression trees | [24, 77, 172] | |
| Random forests | [55] | |
| Linear or quadratic discriminant analysis | [55, 132, 149] | |
| Logistic regression | [55, 148] | |
| Support vector machines | [162, 172] | |
| Artificial neural networks | [55, 172] | |