.Rhistory

# if (!require("hrbrthemes") ) install.packages("hrbrthemes"); library("hrbrthemes");
if (!require("tidyverse") ) install.packages("tidyverse"); library("tidyverse");
if (!require("patchwork") ) install.packages("patchwork"); library("patchwork");
#######################################################################
# R code accompanying Beffara, Bret & Nalborczyk (2019)
# OSF projet: https://osf.io/mwtvk/
# ---------------------------------------------------------
# Sequential analyses procedure, main script
# Written by Ladislas Nalborczyk
# E-mail: ladislas.nalborczyk@gmail.com
# Last update: March 2, 2021
################################################################
#########################################################
# Installing / loading relevant packages
###################################################
# ggplot2 themes
# if (!require("hrbrthemes") ) install.packages("hrbrthemes"); library("hrbrthemes")
# Data formatting, manipulation, and ploting
if (!require("tidyverse") ) install.packages("tidyverse"); library("tidyverse")
# Running tasks on parallel cores
if (!require("parallel") ) install.packages("parallel"); library("parallel")
# Sending emails from R (via gmail)
if (!require("gmailr") ) install.packages("gmailr"); library("gmailr")
# Bayesian multilevel regression models
if (!require("brms") ) install.packages("brms"); library("brms")
# OSF interface
if (!require("osfr") ) {
install.packages("remotes")
remotes::install_github("centerforopenscience/osfr")
library("osfr")
}
?mime
results <- "babar"
email <- # writes email
mime(
To = "brice.beffara@gmail.com",
To = "ameliebret@gmail.com",
To = "ladislas.nalborczyk@gmail.com",
From = "ladislas.nalborczyk@gmail.com",
Subject = "Sequential analysis",
body = results # results of the sequential analysis (stop or continue)
)
?gm_mime
email <- # writes email
gm_mime(
To = "brice.beffara@gmail.com",
To = "ameliebret@gmail.com",
To = "ladislas.nalborczyk@gmail.com",
From = "ladislas.nalborczyk@gmail.com",
Subject = "Sequential analysis",
body = results # results of the sequential analysis (stop or continue)
)
# sends email
send_message(email)
# sends email
gm_send_message(email)
?gm_send_message
?gm_auth_configure
# configure gmail app
gm_auth_configure()
#######################################################################
# R code accompanying Beffara, Bret & Nalborczyk (2019)
# OSF projet: https://osf.io/mwtvk/
# ---------------------------------------------------------
# Sequential analyses procedure, main script
# Written by Ladislas Nalborczyk
# E-mail: ladislas.nalborczyk@gmail.com
# Last update: March 2, 2021
################################################################
#########################################################
# Installing / loading relevant packages
###################################################
# ggplot2 themes
# if (!require("hrbrthemes") ) install.packages("hrbrthemes"); library("hrbrthemes")
# Data formatting, manipulation, and ploting
if (!require("tidyverse") ) install.packages("tidyverse"); library("tidyverse")
# Running tasks on parallel cores
if (!require("parallel") ) install.packages("parallel"); library("parallel")
# Sending emails from R (via gmail)
if (!require("gmailr") ) install.packages("gmailr"); library("gmailr")
# Bayesian multilevel regression models
if (!require("brms") ) install.packages("brms"); library("brms")
# OSF interface
if (!require("osfr") ) {
install.packages("remotes")
remotes::install_github("centerforopenscience/osfr")
library("osfr")
}
#########################################################################
# set working directory to the script's parent folder
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ! temporary solution, only works with RStudio... !
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ---------------------------------------------------------------
# setwd(dirname(rstudioapi::getActiveDocumentContext()$path) )
###################################################################
# Scheduling task (retrieving data and running analyses)
#############################################################
####################################################
# For windows from R (not tested)
# Using the taskscheduleR package
# https://github.com/bnosac/taskscheduleR
###############################################
# if (!require("taskscheduleR") ) install.packages("taskscheduleR")
# library(taskscheduleR)
# get a data.frame of all tasks
# tasks <- taskscheduler_ls(id = "sequential_analysis")
# if it does not exist yet, create automation to run script every hour
# if (is.null(tasks) ) {
# taskscheduler_create(
#     taskname = "sequential_analysis", rscript = "main_script.R",
#     schedule = "HOURLY"
#     )
# }
# delete the task
# taskscheduler_delete(taskname = "sequential_analysis")
# NB: tasks can also be created/deleted from RStudio using the
# taskscheduleR add-in (see the documentation on Github)
##################################################
# For Unix/Linux systems from R
# Using the cronR package
# https://github.com/bnosac/cronR
#########################################
if (!require("cronR") ) remotes::install_github("bnosac/cronR")
library(cronR)
# check whether the task already exists
tasks <- cron_ls(id = "sequential_testing")
# if it does not exist yet, create automation to run script every hour
if (is.null(tasks) ) {
script <- cron_rscript("scripts/main_script.R", log_append = FALSE)
cron_add(script, frequency = "hourly", id = "sequential_testing")
}
# delete the task
# cron_add(script, frequency = "minutely", id = "sequential_testing")
# cron_clear(ask = FALSE)
# NB: tasks can also be created/deleted from RStudio using the
# cronR add-in (see the documentation on Github)
##################################################################
# Retrieving the data files
######################################################
# count the number of datafiles that have already been downloaded
n_data <- list.files(path = "data/", pattern = "csv") %>% length %>% as.numeric
# retrieve OSF data
osf_data <-
# retrieve the OSF project
osf_retrieve_node("mwtvk") %>%
# list components
osf_ls_nodes() %>%
# retrieve the component containing the data
filter(name == "data_emo_stroop_xp") %>%
# list .csv files in this repository
osf_ls_files(pattern = "csv", n_max = Inf) %>%
# remove .csv files ending with "_1"
filter(!str_detect(name, "_1") )
# IF new data is present on OSF, download it, else, quit the script
if (nrow(osf_data) > n_data) {
osf_data %>%
# mutate(path = paste(here("data"), "/", .$name, sep = "") ) %>%
# for each line (each csv file)
{split(., 1:nrow(.) )} %>%
# download it
lapply(osf_download, conflicts = "overwrite")
# download it and put it in the "data" folder
# lapply(osf_download, path = paste(here("data"), .$name, sep = "/ "), overwrite = TRUE)
# lapply(osf_download, path = glue::glue(here("data"), .$name, sep = ""), overwrite = TRUE)
} else {
quit(save = "no")
}
#######################################################################
# R code accompanying Beffara, Bret & Nalborczyk (2019)
# OSF projet: https://osf.io/mwtvk/
# ---------------------------------------------------------
# Sequential analyses procedure, main script
# Written by Ladislas Nalborczyk
# E-mail: ladislas.nalborczyk@gmail.com
# Last update: March 2, 2021
################################################################
#########################################################
# Installing / loading relevant packages
###################################################
# ggplot2 themes
# if (!require("hrbrthemes") ) install.packages("hrbrthemes"); library("hrbrthemes")
# Data formatting, manipulation, and ploting
if (!require("tidyverse") ) install.packages("tidyverse"); library("tidyverse")
# Running tasks on parallel cores
if (!require("parallel") ) install.packages("parallel"); library("parallel")
# Sending emails from R (via gmail)
if (!require("gmailr") ) install.packages("gmailr"); library("gmailr")
# Bayesian multilevel regression models
if (!require("brms") ) install.packages("brms"); library("brms")
# OSF interface
if (!require("osfr") ) {
install.packages("remotes")
remotes::install_github("centerforopenscience/osfr")
library("osfr")
}
#########################################################################
# set working directory to the script's parent folder
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ! temporary solution, only works with RStudio... !
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ---------------------------------------------------------------
# setwd(dirname(rstudioapi::getActiveDocumentContext()$path) )
###################################################################
# Scheduling task (retrieving data and running analyses)
#############################################################
####################################################
# For windows from R (not tested)
# Using the taskscheduleR package
# https://github.com/bnosac/taskscheduleR
###############################################
# if (!require("taskscheduleR") ) install.packages("taskscheduleR")
# library(taskscheduleR)
# get a data.frame of all tasks
# tasks <- taskscheduler_ls(id = "sequential_analysis")
# if it does not exist yet, create automation to run script every hour
# if (is.null(tasks) ) {
# taskscheduler_create(
#     taskname = "sequential_analysis", rscript = "main_script.R",
#     schedule = "HOURLY"
#     )
# }
# delete the task
# taskscheduler_delete(taskname = "sequential_analysis")
# NB: tasks can also be created/deleted from RStudio using the
# taskscheduleR add-in (see the documentation on Github)
##################################################
# For Unix/Linux systems from R
# Using the cronR package
# https://github.com/bnosac/cronR
#########################################
if (!require("cronR") ) remotes::install_github("bnosac/cronR")
library(cronR)
# check whether the task already exists
tasks <- cron_ls(id = "sequential_testing")
# if it does not exist yet, create automation to run script every hour
if (is.null(tasks) ) {
script <- cron_rscript("scripts/main_script.R", log_append = FALSE)
cron_add(script, frequency = "hourly", id = "sequential_testing")
}
# delete the task
# cron_add(script, frequency = "minutely", id = "sequential_testing")
# cron_clear(ask = FALSE)
# NB: tasks can also be created/deleted from RStudio using the
# cronR add-in (see the documentation on Github)
##################################################################
# Retrieving the data files
######################################################
# count the number of datafiles that have already been downloaded
n_data <- list.files(path = "data/", pattern = "csv") %>% length %>% as.numeric
# retrieve OSF data
osf_data <-
# retrieve the OSF project
osf_retrieve_node("mwtvk") %>%
# list components
osf_ls_nodes() %>%
# retrieve the component containing the data
filter(name == "data_emo_stroop_xp") %>%
# list .csv files in this repository
osf_ls_files(pattern = "csv", n_max = Inf) %>%
# remove .csv files ending with "_1"
filter(!str_detect(name, "_1") )
# IF new data is present on OSF, download it, else, quit the script
if (nrow(osf_data) > n_data) {
osf_data %>%
# mutate(path = paste(here("data"), "/", .$name, sep = "") ) %>%
# for each line (each csv file)
{split(., 1:nrow(.) )} %>%
# download it
lapply(osf_download, conflicts = "overwrite")
# download it and put it in the "data" folder
# lapply(osf_download, path = paste(here("data"), .$name, sep = "/ "), overwrite = TRUE)
# lapply(osf_download, path = glue::glue(here("data"), .$name, sep = ""), overwrite = TRUE)
} else {
quit(save = "no")
}
#######################################################################
# R code accompanying Beffara, Bret & Nalborczyk (2019)
# OSF projet: https://osf.io/mwtvk/
# ---------------------------------------------------------
# Sequential analyses procedure, main script
# Written by Ladislas Nalborczyk
# E-mail: ladislas.nalborczyk@gmail.com
# Last update: March 3, 2021
################################################################
#########################################################
# Installing / loading relevant packages
###################################################
# ggplot2 themes
# if (!require("hrbrthemes") ) install.packages("hrbrthemes"); library("hrbrthemes")
# Data formatting, manipulation, and ploting
if (!require("tidyverse") ) install.packages("tidyverse"); library("tidyverse")
# Running tasks on parallel cores
if (!require("parallel") ) install.packages("parallel"); library("parallel")
# Sending emails from R (via gmail)
if (!require("gmailr") ) install.packages("gmailr"); library("gmailr")
# Bayesian multilevel regression models
if (!require("brms") ) install.packages("brms"); library("brms")
# OSF interface
if (!require("osfr") ) {
install.packages("remotes")
remotes::install_github("centerforopenscience/osfr")
library("osfr")
}
#########################################################################
# set working directory to the script's parent folder
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ! temporary solution, only works with RStudio... !
# !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
# ---------------------------------------------------------------
# setwd(dirname(rstudioapi::getActiveDocumentContext()$path) )
###################################################################
# Scheduling task (retrieving data and running analyses)
#############################################################
####################################################
# For windows from R (not tested)
# Using the taskscheduleR package
# https://github.com/bnosac/taskscheduleR
###############################################
# if (!require("taskscheduleR") ) install.packages("taskscheduleR")
# library(taskscheduleR)
# get a data.frame of all tasks
# tasks <- taskscheduler_ls(id = "sequential_analysis")
# if it does not exist yet, create automation to run script every hour
# if (is.null(tasks) ) {
# taskscheduler_create(
#     taskname = "sequential_analysis", rscript = "main_script.R",
#     schedule = "HOURLY"
#     )
# }
# delete the task
# taskscheduler_delete(taskname = "sequential_analysis")
# NB: tasks can also be created/deleted from RStudio using the
# taskscheduleR add-in (see the documentation on Github)
##################################################
# For Unix/Linux systems from R
# Using the cronR package
# https://github.com/bnosac/cronR
#########################################
if (!require("cronR") ) remotes::install_github("bnosac/cronR")
library(cronR)
# check whether the task already exists
tasks <- cron_ls(id = "sequential_testing")
# if it does not exist yet, create automation to run script every hour
if (is.null(tasks) ) {
script <- cron_rscript("scripts/main_script.R", log_append = FALSE)
cron_add(script, frequency = "hourly", id = "sequential_testing")
}
# delete the task
# cron_add(script, frequency = "minutely", id = "sequential_testing")
# cron_clear(ask = FALSE)
# NB: tasks can also be created/deleted from RStudio using the
# cronR add-in (see the documentation on Github)
##################################################################
# Retrieving the data files
######################################################
# count the number of datafiles that have already been downloaded
n_data <- list.files(path = "data/", pattern = "csv") %>% length %>% as.numeric
# retrieve OSF data
osf_data <-
# retrieve the OSF project
osf_retrieve_node("mwtvk") %>%
# list components
osf_ls_nodes() %>%
# retrieve the component containing the data
filter(name == "data_emo_stroop_xp") %>%
# list .csv files in this repository
osf_ls_files(pattern = "csv", n_max = Inf) %>%
# remove .csv files ending with "_1"
filter(!str_detect(name, "_1") )
# IF new data is present on OSF, download it, else, quit the script
if (nrow(osf_data) > n_data) {
osf_data %>%
# mutate(path = paste(here("data"), "/", .$name, sep = "") ) %>%
# for each line (each csv file)
{split(., 1:nrow(.) )} %>%
# download it
lapply(osf_download, conflicts = "overwrite")
# download it and put it in the "data" folder
# lapply(osf_download, path = paste(here("data"), .$name, sep = "/ "), overwrite = TRUE)
# lapply(osf_download, path = glue::glue(here("data"), .$name, sep = ""), overwrite = TRUE)
} else {
# quit(save = "no")
}
###############################################################
# Importing and formatting the data
######################################################
# listing all cvs files
data_files <-
list.files(path = "data/", pattern = "csv") %>%
# converting to dataframe
data.frame(file = ., stringsAsFactors = FALSE) %>%
# removing "deleted" files
filter(!stringr::str_detect(file, "DELETED") ) %>%
# splitting file name by date of acquisition
separate(col = file, into = c("day", "hour"), remove = FALSE, sep = "\\ ") %>%
mutate(day = sub(".*stroop_", "", day), hour = sub(".csv*", "", hour) ) %>%
# sorting files by date of acquisition
arrange(day, hour)
for (i in 1:nrow(data_files) ) {
if (i == 1) data <- NULL
# import data in a dataframe
d <- read.csv(paste0("data/", data_files$file[i]) ) # %>% data.frame
if (nrow(d) == 0) { # if file is empty
next # go to next iteration
} else {
temp_data <-
d %>%
# removing the last (empty) column
select(-ncol(.) ) %>%
# keeping only the relevant columns
select(
stimulus = stim, actor, emotion = emo, word = print_word,
congruency = congr, corr, resp.keys, resp.corr, resp.rt,
participant, date
) %>%
# convert response time to ms
mutate(resp.rt = resp.rt * 1000) %>%
# remove NAs
na.omit
}
if (is.null(data) ) data <- temp_data else data <- rbind(data, temp_data)
}
# remove all files except data
rm(list = setdiff(ls(), "data") )
##########################################################
# Sequential analyses
# ---------------------------------------
# 1) Defining the model of interest
# 2) Fitting it sequentially
################################################
# defining/fitting the model (multilevel ExGaussian model)
model <- brm(
resp.rt ~ 1 + congruency + (1 + congruency | participant),
# exgaussian model
family = exgaussian(),
# defining weakly informative priors (on the scale of the data)
prior = c(
# prior(normal(500, 100), class = Intercept),
prior(normal(0, 10), class = b)
# prior(exponential(0.01), class = sd),
# prior(exponential(0.01), class = sigma)
),
# specifying the dataset
data = data,
# number of chains
chains = 2,
# number of parallel cores
cores = parallel::detectCores(),
# number of iterations and warmup
warmup = 2000, iter = 5000,
control = list(adapt_delta = 0.95),
# sampling from prior (needed to compute the BF)
sample_prior = TRUE
)
# retrieving the sequential_analysis() function
source("scripts/sequential_analyses.R")
# running and storing the results of the sequential analysis (stop or continue)
# see sequential_analyses.R for documentation on the function arguments
results <- sequential_analysis(
model, cleaning = TRUE, type = "SBF",
nmin = 20, step = 1, hypothesis = "congruency = 0",
threshold = 20, rope = c(-0.1, 0.1), precision = 0.16, blind = FALSE
)
# saving the SBF evolution
# save(results, file = "sbf_evolution.rds")
# plotting the evolution
# results %>%
#     ggplot(aes(x = participant, y = 1 / BF) ) +
#     geom_hline(yintercept = 20, linetype = 2) +
#     geom_segment(
#         aes(x = 20, xend = 23, y = 1 / 0.03853322, yend = 1 / 0.03853322),
#         colour = "orangered", linetype = 3
#         ) +
#     geom_segment(
#         aes(x = 23, xend = 23, y = 10, yend = 1 / 0.03853322),
#         colour = "orangered", linetype = 3
#         ) +
#     geom_line() +
#     geom_smooth(colour = "black", method = "lm") +
#     geom_point(pch = 21, fill = "white", size = 2) +
#     hrbrthemes::theme_ipsum_rc() +
#     labs(
#         title = "Evolution of the sequential Bayes factor procedure",
#         subtitle = expression(
#             paste(
#                 "Evolution of the  ", BF[10], " computed from ",
#                 n[min], " = 20 participants to ", n[max],
#                 " = 45 participants", sep = ""
#                 )
#             ),
#         x = "Sample size",
#         y = expression(paste("Bayes factor (", BF[10], ")", sep = "") )
#         )
############################################################
# Sending the results by e-mail (using gmailr)
######################################################
# configure gmail app
# gm_auth_configure()
email <- # writes email
gm_mime(
To = "brice.beffara@gmail.com",
To = "ameliebret@gmail.com",
To = "ladislas.nalborczyk@gmail.com",
From = "ladislas.nalborczyk@gmail.com",
Subject = "Sequential analysis",
body = results # results of the sequential analysis (stop or continue)
)
# sends email
gm_send_message(email)
# quits rstudio
quit(save = "no")