_load-data-explore.qmd

```{r}
#| label: load-data

library(dplyr)
library(tidyr)
library(stringr)
library(purrr)
library(readr)
library(tibble)
library(forcats)
library(glue)
library(jsonlite)
library(redivis)

irw_meta <- redivis$user("bdomingu")$dataset("irw_meta:bdxt")
metadata_table <- irw_meta$table("metadata:h5gs")$to_tibble()

metadata <- metadata_table |>
  mutate(table = str_to_lower(table)) |>
  mutate(variable = str_split(variables, "\\| "),
         prefix = str_extract_all(variables, "(?<= )[A-z_]*?(?=_)") |>
           map(unique) |> map(sort)) |>
         # prefix = if_else(map_int(prefix, length) == 0, list("[no prefix]"), prefix)) |>
  mutate(longitudinal = if_else(longitudinal, "longitudinal", "cross-sectional"),
         longitudinal = as.list(longitudinal)) |>
  filter(n_categories != 0)

var_vals <- metadata |>
  select(table, variable, prefix) |>#, longitudinal) |>
  pivot_longer(cols = -table, names_to = "type", values_to = "value") |>
  unnest(value) |>
  count(type, value) |>
  filter(n >= 10) |>
  arrange(type, desc(n)) |>
  select(-n) |>
  group_by(type) |>
  summarise(vals = list(value)) |>
  deframe()

biblio <- irw_meta$table("biblio:qahg")$to_tibble()
bib_data <- biblio |>
  mutate(table = str_to_lower(table)) |>
  # select(table, license = `Derived_License`)
  select(table, license = `Derived_License`, data_url = URL__for_data_,
         description = Description, reference = Reference_x, doi = DOI__for_paper_)

tag_table <- irw_meta$table("tags:7nkh")$to_tibble()

na_vals <- c("no access to the osf page", "non-verbal task",
             "I can't find the description of this dataset",
             "missing description", "need help", "no link or info",
             "Missing (NA)")

comma <- "~"
tags <- tag_table |>
  mutate(table = str_to_lower(table)) |>
  left_join(bib_data |> select(table, license)) |>
  mutate(across(everything(), \(s) if_else(s %in% na_vals, "NA", s))) |>
  mutate(across(everything(), \(s) replace_na(s, "NA"))) |>
  mutate(sample = sample |> str_replace_all(",(?= etc)", comma) |> str_remove_all('\\"')) |>
  mutate(across(-table, \(s) s |> str_split(",") |> map(str_trim) |> map(\(x) str_replace_all(x, comma, ",")))) |>
  left_join(metadata |> select(table, longitudinal)) |>
  relocate(longitudinal, .before = age_range)

color_vars <- tags |>
  pivot_longer(cols = -table, names_to = "tag", values_to = "value") |>
  distinct(tag, value) |>
  mutate(tag = fct_inorder(tag)) |>
  count(tag) |>
  filter(n <= 10) |>
  pull(tag) |>
  as.character()

sort_alpha <- \(v) v |> fct_relevel("NA", after = Inf) |> levels()
sort_n <- \(v) v |> fct_infreq() |> fct_relevel("NA", after = Inf) |> levels()

age_range_vals <- c("Child (<18y)", "Adult (18+)", "Elderly (minimum age >50)", "Mixed", "Non-human", "NA")
child_age_vals <- c("Early (<6y)", "Child (6-12y)", "Adolescent (12-18y)", "NA")

sort_funs <- list(
  "age_range" = \(v) age_range_vals,
  "child_age__for_child_focused_studies_" = \(v) child_age_vals,
  "construct_name" = sort_alpha,
  "construct_type" = sort_alpha,
  "item_format" = sort_n,
  "license" = sort_n,
  "longitudinal" = sort_n,
  "measurement_tool" = sort_alpha,
  "primary_language_s_" = sort_n,
  "sample" = sort_n
)

sort_tag_values <- \(tag, values) exec(sort_funs[[tag]], values)

tag_vals <- tags |>
  select(-construct_name) |>
  pivot_longer(cols = -table, names_to = "tag", values_to = "value") |>
  unnest(value) |>
  group_by(tag) |>
  summarise(values = list(value)) |>
  mutate(vals = map2(tag, values, sort_tag_values)) |>
  select(-values) |>
  deframe()

ds <- c("item_response_warehouse", "item_response_warehouse_2")
urls <- ds |>
  map(\(d) redivis$user("datapages")$dataset(d)$list_tables() |>
        map(\(t) tibble(table = t$name, url = t$properties$url)) |>
        list_rbind()) |>
  list_rbind() |>
  mutate(table = str_to_lower(table), url = str_remove(url, "\\?.*$"))

datasets <- metadata |> inner_join(tags) |> inner_join(urls) |> inner_join(bib_data |> select(-license)) |> arrange(table)

# save(datasets, var_vals, tag_vals, color_vars, file = "ojs_data.RData")
# load("ojs_data.RData")

ojs_define(datasets = datasets)
ojs_define(tags = tag_vals)
ojs_define(vars = var_vals)
ojs_define(color_vars = color_vars)
```