Cormack-Jolly-Seber (CJS) model using nimbleEcology

cjs_nimbleEcology.Rmd

---
title: "R Notebook"
output: html_notebook
---

# Cormack-Jolly-Seber (CJS) model

```{r setup}
library(AHMbook)
library(nimbleEcology)
```

## Generation of simulated data

Simulated data are generated using the `simOcc` function of the `AHMbook` package.

- Number of occasions: 6
- Number of marked individuals: 20
- Apparent suvival probability : 0.7
- Recapture probability: 0.4

```{r gen_data}
set.seed(123)
cjs_data <- AHMbook::simCJS(
  n.occ = 6,
  n.marked = 20,
  phi = 0.7,
  p = 0.4,
  show.plot = FALSE)
```

## View data

Number of living individuals

```{r view_total}
cjs_data$n.alive
```

Number of individuals captured or recaptured

```{r view_data1}
apply(cjs_data$ch, 2, sum)
```

## Model

```{r model}
mc <- nimbleCode({
  for (n in 1:N) {
    ch[n, F[n]:T] ~ dCJS_ss(phi, p, len = T - F[n] + 1)
  }
  # priors
  phi ~ dunif(0, 1)
  p ~ dunif(0, 1)
})
```

## Fitting

```{r fit}
nimble_const <- list(N = cjs_data$n.ind,
                     T = cjs_data$n.occ,
                     F = cjs_data$f)
nimble_data <- list(ch = cjs_data$ch)
nimble_init <- function() {
  list(phi = runif(1, 0, 1),
       p = runif(1, 0, 1))}
nimble_fit <- nimbleMCMC(code = mc,
                         constants = nimble_const,
                         data = nimble_data,
                         inits = nimble_init,
                         monitors = c("phi", "p"),
                         niter = 8000,
                         nburnin = 4000,
                         nchains = 3,
                         setSeed = 1:3,
                         summary = TRUE,
                         samples = TRUE)
```

### Results

```{r results}
nimble_fit$summary
```