##' Prepare CRTv
##'
##' Template for the functions to prepare specific tasks. Most of this file should not be changed
##' Things to change:
##' - Name of function: prepare_CRTv -> prepare_[value of short_name_scale_str]
##' - dimensions parameter in standardized_names()
##' - 2 [ADAPT] chunks
##'
##' @title prepare_CRTv
##'
##' @param short_name_scale_str
##' @param DF_clean
##'
##' @return
##' @author gorkang
##' @export
prepare_CRTv <- function(DF_clean, short_name_scale_str, output_formats) {
# DEBUG
# targets::tar_load_globals()
# jsPsychHelpeR::debug_function(prepare_CRTv)
# [ADAPT 1/3]: Items to ignore and reverse, dimensions -----------------------
# ****************************************************************************
description_task = "" # Brief description here
items_to_ignore = c("000") # Ignore these items: If nothing to ignore, keep as is
items_to_reverse = c("000") # Reverse these items: If nothing to reverse, keep as is
## NameDimension1, NameDimension2 should be the names of the dimensions
## Inside each c() create a vector of the item numbers for the dimension
## Add lines as needed. If there are no dimensions, keep as is
items_dimensions = list(
NameDimension1 = c("000")
)
# [END ADAPT 1/3]: ***********************************************************
# ****************************************************************************
# Standardized names ------------------------------------------------------
names_list = standardized_names(short_name_scale = short_name_scale_str,
# dimensions = c("NameDimension1", "NameDimension2"), # Use names of dimensions, "" or comment out line
help_names = FALSE) # help_names = FALSE once the script is ready
# Create long -------------------------------------------------------------
DF_long_RAW = create_raw_long(DF_clean, short_name_scale = short_name_scale_str, numeric_responses = FALSE, help_prepare = FALSE)
# Create long DIR ------------------------------------------------------------
DF_long_DIR =
DF_long_RAW |>
dplyr::select(id, trialid, RAW) |>
# [ADAPT]: RAW to DIR for individual items -----------------------------------
# ****************************************************************************
# [REMEMBER]: These regular expressions are most likely WRONG
dplyr::mutate(
DIR =
dplyr::case_when(
grepl("01", trialid) & grepl("mar[i-í]a", RAW, ignore.case = T) ~ 1,
grepl("02", trialid) & (grepl("segundo", RAW, ignore.case = T) | grepl("2", RAW, ignore.case = T) | grepl("dos", RAW, ignore.case = T)) ~ 1,
##No captura alguna: "Se verá cuando mueran"
grepl("03", trialid) & ((grepl("ningun", RAW, ignore.case = T) & !grepl("muerto", RAW, ignore.case = T)) |
grepl("breviv", RAW, ignore.case = T) | grepl("vivo", RAW, ignore.case = T) | grepl("no est[aá]n", RAW, ignore.case = T)) ~ 1,
grepl("04", trialid) & ((!grepl("p.jaro", RAW) & !grepl("ardilla", RAW, ignore.case = T) & !grepl("mono", RAW, ignore.case = T))) ~ 1,
grepl("05", trialid) & grepl("no", RAW, ignore.case = T) ~ 1,
# No captura: Suponiendo que existió Moises y dicha arca, a lo menos 2
grepl("06", trialid) & ((grepl("no", RAW, ignore.case = T) | grepl("ninguno", RAW, ignore.case = T) | (grepl("no[ée]", RAW, ignore.case = T))) |
(!grepl("no se", RAW) & !grepl("", RAW) & !grepl(" ", RAW) & !grepl("[12]", RAW, ignore.case = T) & !grepl("[unodos]", RAW, ignore.case = T))) ~ 1,
grepl("07", trialid) & ((grepl("humo", RAW, ignore.case = T) | grepl("el[e-é]ctrico", RAW, ignore.case = T))) ~ 1,
grepl("08", trialid) & grepl("f[o-ó][so]foro", RAW, ignore.case = T) ~ 1,
grepl("09", trialid) & (!grepl("sí", RAW, ignore.case = T) & (!grepl("por qu[eé] no", RAW, ignore.case = T)) &
(grepl("muerto", RAW, ignore.case = T) | grepl("muri[oó]", RAW, ignore.case = T) |
grepl("imposible", RAW, ignore.case = T) |
(grepl("no", RAW, ignore.case = T) & !grepl("cuñada", RAW, ignore.case = T)))) ~ 1,
grepl("10", trialid) & ((grepl("ninguna", RAW, ignore.case = T) | grepl("amarilla", RAW, ignore.case = T))) ~ 1,
TRUE ~ 0
)
)
# [END ADAPT]: ***************************************************************
# ****************************************************************************
# Create DF_wide_RAW_DIR -----------------------------------------------------
DF_wide_RAW =
DF_long_DIR |>
tidyr::pivot_wider(
names_from = trialid,
values_from = c(RAW, DIR),
names_glue = "{trialid}_{.value}") |>
# NAs for RAW and DIR items
dplyr::mutate(!!names_list$name_RAW_NA := rowSums(is.na(across((-matches(paste0(short_name_scale_str, "_", items_to_ignore, "_RAW")) & matches("_RAW$"))))),
!!names_list$name_DIR_NA := rowSums(is.na(across((-matches(paste0(short_name_scale_str, "_", items_to_ignore, "_DIR")) & matches("_DIR$"))))))
# [ADAPT 3/3]: Scales and dimensions calculations ----------------------------
# ****************************************************************************
DF_wide_RAW_DIR =
DF_wide_RAW |>
dplyr::mutate(
# Score Scale
!!names_list$name_DIRt := rowSums(across(all_of(matches("_DIR$"))), na.rm = TRUE)
)
# [END ADAPT]: ***************************************************************
# ****************************************************************************
# CHECK NAs -------------------------------------------------------------------
check_NAs(DF_wide_RAW_DIR)
# Save files --------------------------------------------------------------
save_files(DF_wide_RAW_DIR, short_name_scale = short_name_scale_str, is_scale = TRUE, output_formats = output_formats)
# Output of function ---------------------------------------------------------
return(DF_wide_RAW_DIR)
}
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.