TidyTuesday, jan14 2025

ttcodeLV.R

# tidytuesday, jan. 14th 2025
library(dplyr)
library(tidyr)
library(stringr)
library(readr)
library(stringi)

conf2024 <- read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2025/2025-01-14/conf2024.csv')
conf2023 <- read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2025/2025-01-14/conf2023.csv')


conf2024 <- 
  conf2024 %>% 
  mutate(speaker_name = str_replace(speaker_name, ", and ",";")) %>%
  mutate(speaker_name = str_replace(speaker_name, " and ",";")) %>%
  separate_rows(speaker_name,sep = ",") |> 
  separate_rows(speaker_name,sep = ";") |> 
  mutate(speaker_name = str_squish(speaker_name)) 

sp23 <- conf2023$speaker_name 
sp24 <- conf2024$speaker_name 

standardize_encoding <- function(vec) {
  # to UTF-8
  vec_utf8 <- enc2utf8(vec)
  # normalize 
  stringi::stri_trans_nfc(vec_utf8)
  
}

sp23 <- standardize_encoding(sp23)
sp24 <- standardize_encoding(sp24)



Speaker <- unique(c(sp23, sp24))

df <- data.frame(
  Speaker = Speaker,
  Event1 = ifelse(Speaker %in% sp23, "Yes", "No"),
  Event2 = ifelse(Speaker %in% sp24, "Yes", "No"),
  Both = ifelse(Speaker %in% sp23 & Speaker %in% sp24, "Yes", "No")
)

firstNameToInitial <- function(names) {
  name_parts <- strsplit(names, " ")
  
  process_name <- function(name) {
    if (length(name) > 1) {
      first_initial <- substr(name[1], 1, 1)
      rest_of_name <- paste(name[-1], collapse = " ")
      paste0(first_initial, ". ", rest_of_name)
    } else {
      name
    }
  }
  
  sapply(name_parts, process_name)
}

df <- 
  df |> mutate(Speaker=firstNameToInitial(Speaker))

df <- 
  df %>% mutate(forwrap=case_when(
    Event1=="Yes"&Event2=="No"~"just1",
    Event1=="No"&Event2=="Yes"~"just2",
    TRUE~"Both"
  ))

df <- df |> mutate(Speaker=str_remove(Speaker,"\\(.*\\)"))
df <- df |> arrange(Speaker)

df$chunk <- cut(1:nrow(df), breaks = 2, labels = c("Chunk 1", "Chunk 2"))


library(ggplot2)
library(patchwork)
library(ggtext)
library(forcats)

dflong <- df |> 
  pivot_longer(cols = c(Event1, Event2)) %>% 
  mutate(name=str_replace(name,"Event1","2023")) %>% 
  mutate(name=str_replace(name,"Event2","2024")) 

# ggplot(aes(x=Speaker,y=name,fill=value))+
#   theme(axis.text.x = element_text(angle = 90))+
#   scale_fill_brewer(guide="none")+
#   facet_wrap(~chunk,ncol=1 )+
#   geom_tile()

dflong <- 
dflong %>% mutate(SpkrHL=
                    if_else(Both=="Yes",paste0("**",Speaker,"**"),Speaker))

dflong |> filter(chunk=="Chunk 1") %>% 
  ggplot(aes(x=fct_inorder(SpkrHL),y=name,fill=value))+
  geom_tile(color="white")+
  theme(text = element_text(family = "FreeSans Regular"),
    axis.text.x = element_markdown(angle = 90,hjust = 1,
                                   size = 7.5,vjust=0.5),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        axis.ticks = element_blank(),
        panel.grid = element_blank(),
        panel.background = element_rect(fill="#D1DBE5"))+
  scale_fill_manual(values = c("transparent","#213D4F"),guide="none")+
  
dflong |> filter(chunk=="Chunk 2") %>% 
  ggplot(aes(x=fct_inorder(SpkrHL),y=name,fill=value))+
  geom_tile(color="white")+
  theme(text = element_text(family = "FreeSans Regular"),
        axis.text.x = element_markdown(angle = 90,hjust = 1,
                                       size = 7.5,vjust=0.5),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        axis.ticks = element_blank(),
        panel.grid = element_blank(),
        panel.background = element_rect(fill="#D1DBE5"))+
  scale_fill_manual(values = c("transparent","#213D4F"),guide="none")+
  plot_layout(ncol=1)+
  plot_annotation(title = "Repeat speakers at posit::conf()",
                  subtitle = "Speakers that presented at both events are highlighted in bold",
                  caption = "TidyTuesday 2025. Luis D. Verde Arregoitia",
                  theme = theme(title = element_text(family = "FreeSans Regular",size = 17)))

ggsave(filename = "tidytuesdayLVjan2025.png",width = 10, height = 4,
       units = "in",dpi = "retina")