Count in-person attendees per talk at RStudio conf.

count-rstudio.R

library(tidyverse)
library(jsonlite)
library(rvest)

raw_json <- fromJSON("https://raw.githubusercontent.com/rstudio/rstudio-conf-2022-program/main/program.json", simplifyVector = FALSE)

raw_df <- raw_json |> 
  enframe() |> 
  unnest_wider(value) |> 
  hoist(talk,
        url = "sched_url")

html_df <- raw_df |> 
  mutate(html = map(url, read_html))
  
get_attendees <- function(html){
  html |> 
    html_element("#sched-page-event-attendees > h2") |> 
    html_text() |> 
    str_extract("[0-9]+") |> 
    as.integer()
}

attendee_df <- html_df |> 
  hoist(
    talk,
    title = "title",
    type = "type", 
    day = "day",
    time_slot = "start_local"
  ) |> 
  hoist(
    session,
    session_slug = "slug"
  ) |> 
  select(-talk, -session, -speakers) |> 
  mutate(attendees = map_int(html, get_attendees)) |> 
  arrange(desc(attendees))

attendee_df

# A tibble: 99 × 9
   name                           url       html       title type  day   time_slot session_slug attendees
   <chr>                          <chr>     <list>     <chr> <chr> <chr> <chr>     <chr>            <int>
 1 applied-machine-learning       https://… <xml_dcmn> Good… Keyn… 2022… 09:00 AM  applied-mac…      1250
 2 collaborate-with-quarto        https://… <xml_dcmn> Hell… Keyn… 2022… 09:00 AM  collaborate…      1188
 3 past-future-shiny              https://… <xml_dcmn> The … Keyn… 2022… 04:30 PM  past-future…      1160
 4 quarto-for-rmarkdown-users     https://… <xml_dcmn> Quar… Regu… 2022… 03:20 PM  rmarkdown-q…       898
 5 new-way-to-build-shiny         https://… <xml_dcmn> A ne… Regu… 2022… 01:30 PM  shiny-app-d…       741
 6 running-shiny-without-server   https://… <xml_dcmn> Runn… Regu… 2022… 11:30 AM  python             726
 7 data-science-training          https://… <xml_dcmn> Data… Keyn… 2022… 04:30 PM  data-scienc…       662
 8 introduction-to-r7             https://… <xml_dcmn> An i… Regu… 2022… 04:00 PM  advanced-r         656
 9 making-data-pipelines-in-r     https://… <xml_dcmn> Maki… Regu… 2022… 01:30 PM  data-quality       608
10 r-python-tableau-love-triangle https://… <xml_dcmn> R, P… Regu… 2022… 11:20 AM  business-in…       602
# … with 89 more rows