Fig. 3
From: Kernel density estimation of conditional distributions to detect responses in satellite tag data
Contours illustrate central region for bivariate density estimate \(f_{S_{31},S_{32}}(u,v)\) for ZcTag093_DUML based on the baseline observation pairs \(\{(s_{31}^{(i)},s_{32}^{(i)})\}\) that summarize 6 hour long “pre/post” windows, with no lag separating them. The green point indicates the CEE pair \((s_{31},s_{32})\), and the green, vertical line illustrates the support for the conditional density \([S_{32}\vert S_{31}]\) used to evaluate \((s_{31},s_{32})\). The distribution’s upper tail consists of all points on the green line above \((s_{31},s_{32})\), and the distribution’s lower tail consists of all points below