Gender Parity Plot

Gender Parity

library(plyr)
library(reshape2)
library(ggplot2)

edx <- read.csv("data//HMXPC13_DI_v2_5-14-14.csv")

#Gender parity by Country

gender.table <- ddply(edx, .(final_cc_cname_DI, gender), summarise, number = length(userid_DI))
levels(gender.table$gender) <- c("none", "female", "male", "other")
gender.table <- dcast(gender.table, final_cc_cname_DI ~ gender, value.var="number")
gender.table$parity <- gender.table$female / gender.table$male

ggplot(gender.table, aes(x=parity, y=reorder(final_cc_cname_DI,parity))) + 
  geom_point(size=5, shape=15, color="orangered") + 
  xlab(expression(frac(Females, Males))) + 
  ylab("Country") 
  
#Grades versus Number of Chapters by Course

edx.nchapters.grade <- edx[,c("nchapters", "grade", "course_id")]
edx.nchapters.grade <- edx.nchapters.grade[complete.cases(edx.nchapters.grade),]

ggplot(edx.nchapters.grade, aes(nchapters, grade)) + 
  geom_point(alpha=1/5, position = "jitter") + 
  geom_smooth(method="lm", color="orangered", size=2) + 
  scale_y_continuous(limits=c(0, 1)) + 
  facet_grid(course_id ~ .) + 
  xlab("Number of chapters with which the Student interacted") + 
  ylab("Grade (>0)")