# 心理学データ解析応用/伴走サイトコード -----------------------------------------------------
#  Programmed by kosugitti
#  Licence ; Creative Commons BY-SA license (CC BY-SA) version 4.0
## Lesson 3. Seven scientist Cmdstanr Version

# 準備 ----------------------------------------------------------------------

rm(list = ls())
library(tidyverse)
library(cmdstanr)
library(bayesplot)
## MAP関数
map_estimation <- function(z) {
  density(z)$x[which.max(density(z)$y)]
}

## MCMCサンプルをデータフレームにする関数
MCMCtoDF <- function(fit) {
  fit$draws() %>%
    posterior::as_draws_df() %>%
    tibble::as_tibble() %>%
    dplyr::select(-lp__, -.draw, -.chain, -.iteration) %>%
    tibble::rowid_to_column("iter") %>%
    tidyr::pivot_longer(-iter) -> MCMCsample
  return(MCMCsample)
}

## MCMCデータフレームを要約する関数
MCMCsummary <- function(MCMCsample) {
  MCMCsample %>%
    dplyr::group_by(name) %>%
    dplyr::summarise(
      EAP = mean(value),
      MED = median(value),
      MAP = map_estimation(value),
      SD = sd(value),
      L95 = quantile(value, prob = 0.025),
      U95 = quantile(value, prob = 0.975),
      Rhat = posterior::rhat_basic(value)
    ) %>%
    mutate(across(where(is.numeric), ~ num(., digits = 3)))
}

# データ ---------------------------------------------------------------------

x <- c(-27.020, 3.570, 8.191, 9.898, 9.603, 9.945, 10.056)

# stanのコンパイル
## cmdstanの場合
library(cmdstanr)
model <- cmdstanr::cmdstan_model("cmdstan/sevenScientist.stan")
model$check_syntax()
fit <- model$sample(
  data = list(Y = x),
)


# トレースプロット ----------------------------------------------------------------

fit$draws() %>% bayesplot::mcmc_trace()

# Stanのオプション指定 ------------------------------------------------------------

fit <- model$sample(
  data = list(Y = x),
  chains = 4,
  parallel_chains = 4,
  iter_warmup = 1000,
  iter_sampling = 5000
)

# 結果の出力 -------------------------------------------------------------------

fit %>%
  MCMCtoDF() %>%
  MCMCsummary()