8.2 Zero-inflated Poisson
squidSim doesn’t generate zero-inflated data directly, but it can be easily made. For example, we can simulate a Poisson variable and a binomial variable, and multiply them together to get our zero-inflated response.
squid_data <- simulate_population(
make_structure(structure = "mother(100)",repeat_obs=5),
n_response = 2,
response_names = c("reproduction","survival"),
parameters=list(
intercept = c(1,0),
mother = list(
vcov = matrix(c(
0.25,0.15,
0.15,1
),nrow=2,ncol=2,byrow=TRUE)
),
residual = list(
vcov = matrix(c(
0.25,0.15,
0.15,1
),nrow = 2,ncol = 2,byrow=TRUE),
beta = matrix(c(
1,0,
0,0
),nrow = 2,ncol = 2,byrow=TRUE)
)
),
family=c("poisson","binomial"),
link=c("log","probit")
)
data <- get_population_data(squid_data)
data$y <- data$survival * data$reproduction
head(data,10)## reproduction survival mother_effect1 mother_effect2 residual1 residual2
## 1 6 0 1.2400171 1.17829778 -0.687284900 0.1006170
## 2 23 1 1.2400171 1.17829778 0.848319584 -1.3063138
## 3 13 1 1.2400171 1.17829778 -0.350498912 -0.9775361
## 4 7 0 1.2400171 1.17829778 -0.872358752 -0.7551186
## 5 7 1 1.2400171 1.17829778 0.005561232 -0.7604861
## 6 10 1 0.3912797 -0.01031889 1.353984498 1.1636901
## 7 4 1 0.3912797 -0.01031889 -0.325840994 0.4764468
## 8 6 1 0.3912797 -0.01031889 -0.108594655 1.4663695
## 9 16 1 0.3912797 -0.01031889 1.264513662 0.7186276
## 10 6 1 0.3912797 -0.01031889 0.293968466 -0.3986855
## mother squid_pop y
## 1 1 1 0
## 2 1 1 23
## 3 1 1 13
## 4 1 1 0
## 5 1 1 7
## 6 2 1 10
## 7 2 1 4
## 8 2 1 6
## 9 2 1 16
## 10 2 1 6plot(table(data$y))