members.only.R

library(dplyr)

df=read.csv("directory.csv.gz")

df$dept2=tolower(df$dept)
df$dept2[grep("psych",df$dept2)]="Psychology"
df$dept2[grep("marketing",df$dept2)]="Marketing"
df$dept2[grep("economics",df$dept2)]="Economics"
df$dept2[df$dept2=="management"]="Management"
df$dept2[df$dept2=="management department"]="Management"
df$dept2[df$dept2=="department of management"]="Management"
df$dept2[df$dept2=="dept of management"]="Management"
df$dept2[grep("management and organizations",df$dept2)]="Management"
df$dept2[df$dept2=="management & organizations dept"]="Management"
df$dept2[grep("organizational",df$dept2)]="Management"
df$dept2[grep("school of business",df$dept2)]="Unspecified B-School"
df$dept2[grep("college of business",df$dept2)]="Unspecified B-School"
df$dept2[grep("school of management",df$dept2)]="Unspecified B-School"
df$dept2[grep("business school",df$dept2)]="Unspecified B-School"
df$dept2[grep("wharton",df$dept2)]="Unspecified B-School"
df$dept2[grep("fuqua",df$dept2)]="Unspecified B-School"
df$dept2[grep("decision",df$dept2)]="Decision Making"
df$dept2[grep("accounting",df$dept2)]="Accounting"
df$dept2[grep("health",df$dept2)]="Health"
df$dept2[grep("medicine",df$dept2)]="Health"

plot_data = df %>% 
  group_by(dept2) %>% 
  summarize(count=length(dept2)) %>% 
  arrange(-count) %>%
  filter(count>=5 & dept2!=" ")

names(plot_data)=c("Department","Count")
plot_data$Department=factor(plot_data$Department,levels=plot_data$Department)

p=ggplot(plot_data,aes(x=Department,y=Count,fill=Department))
p=p+geom_bar(stat="identity") 
p=p+labs(x="",y="",title="Members in Areas with 5+ Members")
p=p+theme(legend.position = "none")
p=p+theme( axis.text.x = element_text(angle=90, vjust=0.5),legend.position="none")
ggsave(plot=p,file="MembersByArea.png",width=4,height=4)