init_list_4 <- list(
trans_xi_level = trans_xi_level_init + 0.0,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1
)
seed_value <- 87123
# -------------------  Carry Out Analysis ------------------
#  Model file
model_file <- file.path(stan_dir, paste0(model_str, ".stan"))
# Output Files
draws_file <- file.path(application_dir, paste0("CSV/",application_str,".",model_str,".Draws."))
output_file <- file.path(application_dir, paste0("Output/",application_str,".",model_str))
# Compile model
mod <- cmdstan_model(model_file)
# # MCMC Draws model
fit <- mod$sample(
data = data_list,
iter_warmup = 5000,
iter_sampling = 40000,
init=list(
init_list_1, # chain 1
init_list_2, # chain 2
init_list_3, # chain 3
init_list_4  # chain 4
),
refresh = 5000,
chains = 4,
parallel_chains = 4,
seed = seed_value
)
# fit diagnostics
sink(paste0(output_file,".diagonstics.txt"))
fit$diagnostic_summary()
# fit summary
sink(paste0(output_file,".summary.txt"))
fit$print(max_rows = 1000)
sink(file = NULL)
# # Save Draws of key parameters
# draws_arr <- fit$draws()
# tmp <- subset_draws(draws_arr, c("xi"))
# xi_draws <- merge_chains(tmp)
# write.csv(xi_draws, paste0(draws_file, "xi.csv"))
# #
# tmp <- subset_draws(draws_arr, c("sigma"))
# sigma_draws <- merge_chains(tmp)
# write.csv(sigma_draws, paste0(draws_file, "sigma.csv"))
# #
# tmp <- subset_draws(draws_arr, c("mu"))
# mu_draws <- merge_chains(tmp)
# write.csv(mu_draws, paste0(draws_file, "mu.csv"))
#  begin by clearing memory
rm(list=ls())
#  Load libraries
library(cmdstanr)
library(posterior)
library(readxl)
# Set directory
setwd("/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/FirmSize")
#
# Specify model input
# --------------- City size Model -------------
stan_dir <- "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Stan_Code"
application_str = "FirmSize"
application_dir = "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/FirmSize"
data_str <- "FirmSize_Data_30"
model_str <- "GEV_4RW_Model"
# Read Data From Excel File
data_file <- file.path(application_dir, paste0(data_str,".xlsx"))
cs_data <- read_excel(data_file,col_names = FALSE)
# Process data and get dimensions
# Convert to matrix
y <- as.matrix(cs_data)
#  Get dimensions of cs_data
T <- dim(y)[1]
k <- dim(y)[2]
# Parameter values
#  xi bounds
x_min <- -0.9
x_max <- 2.1
# Scale gamma parameters
sg_xi <- 1.0
sg_alpha <- 0.5
sg_s <- 0.5
# Create list of data inputs for Stan
data_list <- list(
T = T,
nobs = k,
xi_min = x_min,
xi_max = x_max,
sg_xi = sg_xi,
sg_alpha = sg_alpha,
sg_s = sg_s,
y = t(y)
)
#  Inittial Values of parameters and seed value
trans_xi_level_init <- 0.0
ln_s_level_init <- 0.0
m_level_init <- 1.0
g_xi_init <- 0.2
g_alpha_init <- 0.2
g_s_init <- 0.2
g_m_init <- 0.2
init_list_1 <- list(
trans_xi_level = trans_xi_level_init,
ln_s_level = ln_s_level_init,
m_level = m_level_init,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_2 <- list(
trans_xi_level = trans_xi_level_init + 0.1,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_3 <- list(
trans_xi_level = trans_xi_level_init - 0.1,
ln_s_level = ln_s_level_init - 0.1,
m_level = m_level_init - 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_4 <- list(
trans_xi_level = trans_xi_level_init + 0.0,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
seed_value <- 87123
#  Model file
model_file <- file.path(stan_dir, paste0(model_str, ".stan"))
# Output Files
draws_file <- file.path(application_dir, paste0("CSV/",application_str,".",model_str,".Draws."))
output_file <- file.path(application_dir, paste0("Output/",application_str,".",model_str))
# Compile model
mod <- cmdstan_model(model_file)
# MCMC Draws model
fit <- mod$sample(
data = data_list,
iter_warmup = 5000,
iter_sampling = 25000,
init=list(
init_list_1, # chain 1
init_list_2, # chain 2
init_list_3, # chain 3
init_list_4  # chain 4
),
refresh = 5000,
chains = 4,
parallel_chains = 4,
seed = seed_value
)
# fit diagnostics
sink(paste0(output_file,".diagonstics.txt"))
fit$diagnostic_summary()
#  fit summary
sink(paste0(output_file,".summary.txt"))
fit$print(max_rows = 10000,variables = c("gamma_xi","gamma_alpha","gamma_s","gamma_m","m_level","ln_s_level","xi","alpha","mu","sigma"))
sink(file = NULL)
# Save Draws of key parameters
draws_arr <- fit$draws()
tmp <- subset_draws(draws_arr, c("gamma_xi"))
gamma_xi_draws <- merge_chains(tmp)
write.csv(gamma_xi_draws, paste0(draws_file, "gamma_xi.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_alpha"))
gamma_alpha_draws <- merge_chains(tmp)
write.csv(gamma_alpha_draws, paste0(draws_file, "gamma_alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_s"))
gamma_s_draws <- merge_chains(tmp)
write.csv(gamma_s_draws, paste0(draws_file, "gamma_s.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_m"))
gamma_m_draws <- merge_chains(tmp)
write.csv(gamma_m_draws, paste0(draws_file, "gamma_m.csv"))
#
tmp <- subset_draws(draws_arr, c("xi"))
xi_draws <- merge_chains(tmp)
write.csv(xi_draws, paste0(draws_file, "xi.csv"))
#
tmp <- subset_draws(draws_arr, c("alpha"))
alpha_draws <- merge_chains(tmp)
write.csv(alpha_draws, paste0(draws_file, "alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("sigma"))
sigma_draws <- merge_chains(tmp)
write.csv(sigma_draws, paste0(draws_file, "sigma.csv"))
#
tmp <- subset_draws(draws_arr, c("mu"))
mu_draws <- merge_chains(tmp)
write.csv(mu_draws, paste0(draws_file, "mu.csv"))
#  begin by clearing memory
rm(list=ls())
#  Load libraries
library(cmdstanr)
library(posterior)
library(readxl)
# Set directory
setwd("/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Returns")
# --- Run for largest returns then for smallest returns
# Specify model input
stan_dir <- "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Stan_Code"
application_dir = "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Returns"
model_str <- "GEV_4RW_Model"
#  Returns Large
application_str = "Returns_Large"
data_str <- "Returns_Large"
#  Returns Small
# application_str = "Returns_Small"
# data_str <- "Returns_Small"
# Read Data From Excel File
data_file <- file.path(application_dir, paste0(data_str,".xlsx"))
ret <- read_excel(data_file,col_names = FALSE)
# Process data and get dimensions
# Convert to matrix
y <- as.matrix(ret)
#  Get dimensions of cs_data
T <- dim(y)[1]
k <- dim(y)[2]
# Parameter values
#  xi bounds
x_min <- -0.9
x_max <- 2.1
# Scale gamma parameters
sg_xi <- 1.0
sg_alpha <- 0.5
sg_s <- 0.5
# Create list of data inputs for Stan
data_list <- list(
T = T,
nobs = k,
xi_min = x_min,
xi_max = x_max,
sg_xi = sg_xi,
sg_alpha = sg_alpha,
sg_s = sg_s,
y = t(y)
)
#  Inittial Values of parameters and seed value
trans_xi_level_init <- 0.0
ln_s_level_init <- -0.5
m_level_init <- 3.0
g_xi_init <- 0.2
g_alpha_init <- 0.2
g_s_init <- 0.2
g_m_init <- 0.2
init_list_1 <- list(
trans_xi_level = trans_xi_level_init,
ln_s_level = ln_s_level_init,
m_level = m_level_init,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_2 <- list(
trans_xi_level = trans_xi_level_init + 0.1,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_3 <- list(
trans_xi_level = trans_xi_level_init - 0.1,
ln_s_level = ln_s_level_init - 0.1,
m_level = m_level_init - 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_4 <- list(
trans_xi_level = trans_xi_level_init + 0.0,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
seed_value <- 87123
#  Model file
model_file <- file.path(stan_dir, paste0(model_str, ".stan"))
# Output Files
draws_file <- file.path(application_dir, paste0("CSV/",application_str,".",model_str,".Draws."))
output_file <- file.path(application_dir, paste0("Output/",application_str,".",model_str))
# Compile model
mod <- cmdstan_model(model_file)
# MCMC Draws model
fit <- mod$sample(
data = data_list,
iter_warmup = 5000,
iter_sampling = 10000,
init=list(
init_list_1, # chain 1
init_list_2, # chain 2
init_list_3, # chain 3
init_list_4  # chain 4
),
refresh = 5000,
chains = 4,
parallel_chains = 4,
seed = seed_value
)
# fit diagnostics
sink(paste0(output_file,".diagonstics.txt"))
fit$diagnostic_summary()
#  fit summary
sink(paste0(output_file,".summary.txt"))
fit$print(max_rows = 10000,variables = c("gamma_xi","gamma_alpha","gamma_s","gamma_m","m_level","ln_s_level","xi","alpha","mu","sigma"))
sink(file = NULL)
# Save Draws of key parameters
draws_arr <- fit$draws()
tmp <- subset_draws(draws_arr, c("gamma_xi"))
gamma_xi_draws <- merge_chains(tmp)
write.csv(gamma_xi_draws, paste0(draws_file, "gamma_xi.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_alpha"))
gamma_alpha_draws <- merge_chains(tmp)
write.csv(gamma_alpha_draws, paste0(draws_file, "gamma_alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_s"))
gamma_s_draws <- merge_chains(tmp)
write.csv(gamma_s_draws, paste0(draws_file, "gamma_s.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_m"))
gamma_m_draws <- merge_chains(tmp)
write.csv(gamma_m_draws, paste0(draws_file, "gamma_m.csv"))
#
tmp <- subset_draws(draws_arr, c("xi"))
xi_draws <- merge_chains(tmp)
write.csv(xi_draws, paste0(draws_file, "xi.csv"))
#
tmp <- subset_draws(draws_arr, c("alpha"))
alpha_draws <- merge_chains(tmp)
write.csv(alpha_draws, paste0(draws_file, "alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("sigma"))
sigma_draws <- merge_chains(tmp)
write.csv(sigma_draws, paste0(draws_file, "sigma.csv"))
#
tmp <- subset_draws(draws_arr, c("mu"))
mu_draws <- merge_chains(tmp)
write.csv(mu_draws, paste0(draws_file, "mu.csv"))
#  begin by clearing memory
rm(list=ls())
#  Load libraries
library(cmdstanr)
library(posterior)
library(readxl)
# Set directory
setwd("/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Returns")
# --- Run for largest returns then for smallest returns
# Specify model input
stan_dir <- "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Stan_Code"
application_dir = "/Users/mwatson/Dropbox/TVExtreme/ReplicationFiles/Stan/Returns"
model_str <- "GEV_4RW_Model"
#  Returns Large
# application_str = "Returns_Large"
# data_str <- "Returns_Large"
#  Returns Small
application_str = "Returns_Small"
data_str <- "Returns_Small"
# Read Data From Excel File
data_file <- file.path(application_dir, paste0(data_str,".xlsx"))
ret <- read_excel(data_file,col_names = FALSE)
# Process data and get dimensions
# Convert to matrix
y <- as.matrix(ret)
#  Get dimensions of cs_data
T <- dim(y)[1]
k <- dim(y)[2]
# Parameter values
#  xi bounds
x_min <- -0.9
x_max <- 2.1
# Scale gamma parameters
sg_xi <- 1.0
sg_alpha <- 0.5
sg_s <- 0.5
# Create list of data inputs for Stan
data_list <- list(
T = T,
nobs = k,
xi_min = x_min,
xi_max = x_max,
sg_xi = sg_xi,
sg_alpha = sg_alpha,
sg_s = sg_s,
y = t(y)
)
#  Inittial Values of parameters and seed value
trans_xi_level_init <- 0.0
ln_s_level_init <- -0.5
m_level_init <- 3.0
g_xi_init <- 0.2
g_alpha_init <- 0.2
g_s_init <- 0.2
g_m_init <- 0.2
init_list_1 <- list(
trans_xi_level = trans_xi_level_init,
ln_s_level = ln_s_level_init,
m_level = m_level_init,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_2 <- list(
trans_xi_level = trans_xi_level_init + 0.1,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_3 <- list(
trans_xi_level = trans_xi_level_init - 0.1,
ln_s_level = ln_s_level_init - 0.1,
m_level = m_level_init - 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
init_list_4 <- list(
trans_xi_level = trans_xi_level_init + 0.0,
ln_s_level = ln_s_level_init + 0.1,
m_level = m_level_init + 0.1,
g_xi = g_xi_init,
g_alpha = g_alpha_init,
g_s = g_s_init,
g_m = g_m_init
)
seed_value <- 87123
#  Model file
model_file <- file.path(stan_dir, paste0(model_str, ".stan"))
# Output Files
draws_file <- file.path(application_dir, paste0("CSV/",application_str,".",model_str,".Draws."))
output_file <- file.path(application_dir, paste0("Output/",application_str,".",model_str))
# Compile model
mod <- cmdstan_model(model_file)
# MCMC Draws model
fit <- mod$sample(
data = data_list,
iter_warmup = 5000,
iter_sampling = 10000,
init=list(
init_list_1, # chain 1
init_list_2, # chain 2
init_list_3, # chain 3
init_list_4  # chain 4
),
refresh = 5000,
chains = 4,
parallel_chains = 4,
seed = seed_value
)
# fit diagnostics
sink(paste0(output_file,".diagonstics.txt"))
fit$diagnostic_summary()
#  fit summary
sink(paste0(output_file,".summary.txt"))
fit$print(max_rows = 10000,variables = c("gamma_xi","gamma_alpha","gamma_s","gamma_m","m_level","ln_s_level","xi","alpha","mu","sigma"))
sink(file = NULL)
# Save Draws of key parameters
draws_arr <- fit$draws()
tmp <- subset_draws(draws_arr, c("gamma_xi"))
gamma_xi_draws <- merge_chains(tmp)
write.csv(gamma_xi_draws, paste0(draws_file, "gamma_xi.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_alpha"))
gamma_alpha_draws <- merge_chains(tmp)
write.csv(gamma_alpha_draws, paste0(draws_file, "gamma_alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_s"))
gamma_s_draws <- merge_chains(tmp)
write.csv(gamma_s_draws, paste0(draws_file, "gamma_s.csv"))
#
tmp <- subset_draws(draws_arr, c("gamma_m"))
gamma_m_draws <- merge_chains(tmp)
write.csv(gamma_m_draws, paste0(draws_file, "gamma_m.csv"))
#
tmp <- subset_draws(draws_arr, c("xi"))
xi_draws <- merge_chains(tmp)
write.csv(xi_draws, paste0(draws_file, "xi.csv"))
#
tmp <- subset_draws(draws_arr, c("alpha"))
alpha_draws <- merge_chains(tmp)
write.csv(alpha_draws, paste0(draws_file, "alpha.csv"))
#
tmp <- subset_draws(draws_arr, c("sigma"))
sigma_draws <- merge_chains(tmp)
write.csv(sigma_draws, paste0(draws_file, "sigma.csv"))
#
tmp <- subset_draws(draws_arr, c("mu"))
mu_draws <- merge_chains(tmp)
write.csv(mu_draws, paste0(draws_file, "mu.csv"))
