Naive Occupancy Estimator¶
-
class
ollin.estimation.occupancy.naive.Model[source]¶ Naive Occupancy Estimator.
The naive model uses the percentage of cameras with some detection to estimate occupancy. Detectability is then defined as the mean proportion of detection at cameras with some detection.
Hence if detection data is:
\[\begin{split}D = \left(d_{ij}\right)_{\substack{1 \leq i \leq N \\ 1 \leq j \leq M}}\end{split}\]Where:
\[\begin{split}d_{ij} = \begin{cases} 1 & \text{Camera } i \text{ made a detection at step } j \\ 0 & \text{Otherwise} \end{cases}\end{split}\]Then, if \(\hat{o_{}}\) and \(\hat{d_{}}\) is detectability:
\[\begin{split}\begin{align} \hat{o_i} & = \max\{d_{ij} \mid 1 \leq j \leq M\} \\ \hat{o_{}} & = \frac{1}{N} \sum_{i = 1}^N \hat{d_i} \end{align}\end{split}\]And
\[\begin{split}\begin{align} \hat{d_i} &= \frac{1}{M} \sum_{j = 1}^M d_{ij} \\ \hat{d_{}} &= \frac{\sum_{i=1}^N \hat{d_i}}{\sum_{i=1}^N \hat{o_i}} \end{align}\end{split}\]
