ito4303

library(nimble)
library(coda)

## Data
k <- 10
x <- c(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
y <- c(1, 2, 2, 6, 4, 5, 8, 9, 9, 9, 10)
n <- length(x)

## Code

ito4303

library(nimble)
library(dplyr)

# Data
set.seed(123)
n <- 40
omega <- 0.4
lambda <- 3
y <- rbinom(n, 1, omega) * rpois(n, lambda)

ito4303

library(rstan)
options(mc.cores = parallel::detectCores())
rstan_options(auto_write = TRUE)

##

set.seed(20180816)
N <- 100
Y <- rnorm(N, 4, 1)

ito4303

# KFAS model with offset

set.seed(20180830)
T <- 100
sigma <- 0.05
dens <- rep(NA, T)
dens[1] <- 0.5
for (t in 2:T)
  dens[t] <- exp(rnorm(1, log(dens[t - 1]), sigma))
area <- rep(c(10, 20), T / 2)

ito4303

---
title: "fit negative binomial to Poisson data"
output: html_notebook
---

## Data

```{r}
library(ggplot2)
library(magrittr)

ito4303

library(sf)
library(jpndistrict)
library(stringr)
library(ggplot2)

# 札幌市
sapporo <- jpn_cities(jis_code = sprintf("%05d", 1101:1110))
ggplot(sapporo) +
  geom_sf(aes(fill = city_code)) +
  scale_fill_discrete(name = "区",

ito4303

#
# Estimating "wiggly" covariate relationships using Gaussian Process
#
# cf. Section 10.14 of
# Kéry and Royle (2016) "Applied Hierarchical Modeling in Ecology" Vol.1
# 

library(rstan)
options(mc.cores = parallel::detectCores())
rstan_options(auto_write = TRUE)

ito4303

data {
  int<lower = 1> M;
  int<lower = 1> J;
  int<lower = 0, upper = 1> Y[M, J];
  vector[M] VegHt;
  matrix[M, J] Wind;
  vector[M] XvegHt;
  vector[M] Xwind;
}
transformed data {

ito4303

data {
  int<lower = 1> N_ind;                     // Number of individuals
  int<lower = 1> N_z;                       // Number of augment observed data
  vector<lower = 0>[N_ind] X;               // Observed distance
  int<lower = 0, upper = 1> Y[N_ind + N_z]; // Augumented inds. have y=0 by
                                            // definition
  real B;                                   // Strip half-width
                                            // Larger than max observed distance
}
parameters {

ito4303

data {
  int<lower = 1> N_ind;                     // Number of individuals
  int<lower = 1> N_z;                       // Number of augment observed data
  int<lower = 0, upper = 1> Y[N_ind + N_z]; // Augumented inds. have y=0 by
                                            // definition
  real B;                                   // Strip half-width
                                            // Larger than max observed distance
  int<lower = 1> N_D;                         // N intervals
  int<lower = 1, upper = N_D> D_class[N_ind]; // Distance class
  real<lower = 0, upper = B> Delta;           // Width of distance bins