Unlocking the power of data visualization with R

library(tidyverse)

paygap <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2022/2022-06-28/paygap.csv')

paygap%>%head

paygap%>%names

df <- paygap%>% # count(employer_id,sort=T)
  arrange(employer_id) %>%
  group_by(employer_id)%>%
  summarize(employer_size,across(.cols = is.numeric,median)) %>%
  ungroup()
  
df

df_bonus

df_bonus <- df%>% #DataExplorer::profile_missing()
  select(employer_id,employer_size,contains("bonus")) %>%
  filter(!employer_size=="Not Provided") %>%
  mutate(id=case_when(employer_size=="Less than 250"~1,
                      employer_size=="250 to 499"~2,
                      employer_size=="500 to 999"~3,
                      employer_size=="1000 to 4999"~4,
                      employer_size=="5000 to 19,999"~5,
                      employer_size=="20,000 or more"~6)) %>%
  relocate(id) %>%
  arrange(id)

df_bonus %>% DataExplorer::profile_missing()

df_bonus %>%
  #group_by(employer_size) %>%
  #summarize(across(is.numeric,.fns = median)) %>%
  #ungroup() %>%
  ggplot(aes(x=male_bonus_percent,y=female_bonus_percent))+
  geom_boxplot(aes(group=employer_size),size=0.2)+ 
  geom_point(size=0.1,aes(color=employer_id))+
  geom_abline(intercept = max(df_bonus$male_bonus_percent),slope=-1,size=0.1,linetype="dashed")+
  geom_abline(intercept = 0,slope=1,size=0.5,linetype="dashed",color="red")+
  geom_smooth(method="lm")

df_bonus%>%
  filter(male_bonus_percent>50,female_bonus_percent<50) %>%
    ggplot(aes(x=male_bonus_percent,y=female_bonus_percent))+
 # geom_boxplot(aes(group=employer_size),size=0.2)+ 
  geom_point(size=0.1,aes(color=employer_id))+
  geom_abline(intercept = max(df_bonus$male_bonus_percent),slope=-1,size=0.5,linetype="dashed")+
  geom_abline(intercept = 0,slope=1,size=0.5,linetype="dashed",color="red")+
  geom_text(aes(label=employer_id),check_overlap = T,size=2)+
  geom_smooth(method="lm")

library(tidymodels)

pay_bonus_pca<- df_bonus %>%
  select(-employer_size) %>%
  drop_na() %>%
  prcomp(scale = TRUE)

tidy(pay_bonus_pca, matrix = "scores") %>%
  filter(PC%in%c(1,2)) %>%
  pivot_wider(names_from=PC,values_from=value)%>%
  janitor::clean_names() %>%
  ggplot(aes(x=x1,y=x2))+
    geom_point()

df%>%count(employer_size)

df %>%
  mutate(id=case_when(employer_size=="Less than 250"~1,
                      employer_size=="250 to 499"~2,
                      employer_size=="500 to 999"~3,
                      employer_size=="1000 to 4999"~4,
                      employer_size=="5000 to 19,999"~5,
                      employer_size=="20,000 or more"~6)) %>%
  relocate(id) %>%
  arrange(id) 
  #DataExplorer::profile_missing()

Sampling

df1<-df %>%
  mutate(id=case_when(employer_size=="Less than 250"~1,
                      employer_size=="250 to 499"~2,
                      employer_size=="500 to 999"~3,
                      employer_size=="1000 to 4999"~4,
                      employer_size=="5000 to 19,999"~5,
                      employer_size=="20,000 or more"~6)) %>%
  relocate(id) %>%
  arrange(id)
df1
# library(sampling)
# strata(df1,size=300)

df2<-df %>% #DataExplorer::profile_missing()
  filter(!is.na(female_lower_quartile),!employer_size=="Not Provided") %>%
  select(!contains("bonus")) %>%
  mutate(id=case_when(employer_size=="Less than 250"~1,
                      employer_size=="250 to 499"~2,
                      employer_size=="500 to 999"~3,
                      employer_size=="1000 to 4999"~4,
                      employer_size=="5000 to 19,999"~5,
                      employer_size=="20,000 or more"~6)) %>%
  relocate(id) %>%
  arrange(id) %>%
  group_by(id,employer_size)%>%
  mutate(n=n()) %>%
  ungroup() %>%
  mutate(pct=round(n/sqrt(sum(n))*100,5))%>%#count(pct,n) %>%
  relocate(n,pct)%>%
  group_by(id,employer_size)%>%
  summarise(across(.cols = everything(),.fns = mean)) %>%
  ungroup() %>%
 # select(-diff_mean_bonus_percent) %>%
  #mutate(diff_mean_bonus=round(male_bonus_percent-female_bonus_percent,2)) %>%
  mutate_if(is.numeric, round, 2)
df2

df2 %>%
  arrange(id)%>%names

df2 %>%
  arrange(id)%>%
  ggplot(aes(x=25,y=fct_reorder(employer_size,id),label=employer_size))+
  geom_point(color="grey80",shape=".")+
  geom_text(hjust=1)+
  geom_text(aes(x=30,label=paste0(diff_mean_hourly_percent,"%"))) +
  # lower
  geomtextpath::geom_textsegment(label="Lower",size=4,
                                 aes(x=female_lower_quartile,xend=male_lower_quartile,yend=employer_size),
               position = position_nudge(y = 0.25),
               color="grey50",size=0.5)+
  geom_point(aes(x=female_lower_quartile,y=employer_size),
             position = position_nudge(y = 0.25),
             color="violet",size=5)+
  geom_point(aes(x=male_lower_quartile,y=employer_size),
             position = position_nudge(y = 0.25),
             color="navy",size=5)+
  # upper
    geomtextpath::geom_textsegment(label="Upper",size=4,
                            aes(x=female_top_quartile,xend=male_top_quartile,yend=employer_size),
                            color="grey50",size=0.5)+
  geom_point(aes(x=female_top_quartile,y=employer_size),color="violet",size=5)+
  geom_point(aes(x=male_top_quartile,y=employer_size),color="navy",size=5)+
  geom_vline(xintercept = 50,size=0.3,linetype="dashed")+

  #scale_x_continuous(limits = c(0,65),breaks = c(40,50,60))+
  labs(y="")+
  ggthemes::theme_fivethirtyeight()+
  theme(axis.text.y = element_blank(),
        panel.grid.major.y = element_blank(),
        panel.grid.major.x = element_blank())

df2 %>%
  arrange(id)%>%
  ggplot(aes(x=25,y=fct_reorder(employer_size,id),label=employer_size))+
  geom_text(hjust=1,fontface="bold",color="#003d59",size=6)+
  geom_text(aes(x=30,label=paste0(diff_median_hourly_percent,"%")),
            fontface="bold",size=4,
            color="#003d59") +
  # lower
  geom_segment(size=3.5,
               aes(x=female_lower_quartile,xend=female_top_quartile,yend=employer_size),
               position = position_nudge(y = 0.2),
               color="#a8bd3a",size=0.5)+
  geom_point(aes(x=female_lower_quartile,y=employer_size),
             position = position_nudge(y = 0.2),
             shape=21,stroke=1,fill="violet",
             color="#a8bd3a",size=5)+
  geom_point(aes(x=female_top_quartile,y=employer_size),
             position = position_nudge(y = 0.2),
             shape=21,stroke=1,fill="violet",
             color="#a8bd3a",size=5)+
  # upper
  geom_segment(size=3.5,
               aes(x=male_lower_quartile,xend=male_top_quartile,yend=employer_size),
               position = position_nudge(y = -0.2),
               color="#a8bd3a",size=0.5)+
  geom_point(aes(x=male_lower_quartile,y=employer_size),
             position = position_nudge(y = -0.2),
             shape=21,stroke=1,fill="navy",
             color="#a8bd3a",size=5)+
  geom_point(aes(x=male_top_quartile,y=employer_size),
             position = position_nudge(y = -0.2),
             shape=21,stroke=1,fill="navy",
             color="#a8bd3a",size=5)+
  # vertical line
  geom_vline(xintercept = c(40,50,60),size=(c(40,100,40)/100),linetype="dashed",color="grey40")+
  scale_x_continuous(limits = c(10,65),breaks = c(40,50,60),labels = paste0(c(40,50,60),"K"))+
  coord_cartesian(clip = "off",ylim = c(1,6))+
  labs(title="Difference in gender PayGap by size of Employers\n\n",
       caption="\nDataSource: #TidyTuesday 2022 week26 ons.gov.uk | DataViz: Federica Gazzelloni (@fgazzelloni)")+
   annotate("text",
            x= c(18,30,50),
            y = c(6.8,6.8,6.8),
           label = c("Employer size", "PayGap", "Gender upper/lower values "),
           family = "", fontface = 3, size=5,color="#a8bd3a") +
    ggthemes::theme_fivethirtyeight()+
  theme(text = element_text(color="#a8bd3a"),
        plot.title = element_text(size=18),
        plot.caption = element_text(size=8.5,family="",face=3),
        axis.text.y = element_blank(),
        axis.text.x = element_text(size=12,face="bold"),
        panel.grid.major.y = element_line(size=14),
        panel.grid.major.x = element_blank(),
        plot.background = element_rect(color="#003d59",fill="#003d59"),
        panel.background = element_rect(color="#003d59",fill="#003d59"))+
     annotate("text",
            x= 20.8,
            y = 0.2,
           label = "PayGap are median values calculated on unique employer id\nfrom hourly pay(%) mean difference\nPink = Female | Blue = Male",
           family = "", fontface = 3, size=2.8,color="white")

ggsave("w26_paygap.png",dpi=320)