library(tidyverse)
library(ggplot2)

Orig <- read_csv("HistoricalEval03-longform.csv")
Orig

# Turn Class into Factor
Orig <- Orig %>% mutate(Class_Levels=as.factor(Class))
Orig

# Change factor labels for plotting porpose
Orig <- Orig %>% 
  mutate(Class_Levels=recode(Class_Levels, 
                    '1'="200 level",
                    '2'="401+461",
                    '3'="4/5/600"  ))

# Combine 2007 and 2007.5 as 2007. 
Orig <- Orig %>% mutate(Semester=round(Semester,0))
Orig

# Use aggregate for cond mean line
Orig_agg <- aggregate(Rating ~ Semester + Class_Levels, data=Orig, mean)
Orig_agg <- as.tibble(Orig_agg)
Orig_agg



# https://www.marsja.se/how-to-add-a-column-to-dataframe-in-r-with-tibble-dplyr/
# https://suzan.rbind.io/2018/02/dplyr-tutorial-2/


# Scatterplot with smoothed cond mean
ggplot(Orig, aes(x=Semester, y=Rating, color=Class_Levels)) + 
  geom_point() + geom_smooth(se=FALSE) + 
  theme(legend.position="right")


ggplot(Orig_agg, aes(x=Semester, y=Rating, color=Class_Levels)) + 
  geom_point() + geom_line() + 
  theme(legend.position="right") +
  ggtitle("Av Rating for each Class_Levels")

# Cond mean only
ggplot(Orig, aes(x=Semester, y=Rating, color=Class_Levels)) + 
  geom_point() + 
  geom_point(data=Orig_agg) + 
  geom_line(data=Orig_agg,lwd=1) + 
  theme(legend.position="right") +
  ggtitle("Cond mean of #19 Rating for each class offered 2007-2017")


# Both combined 
ggplot(Orig, aes(x=Semester, y=Rating, color=Class_Levels)) + 
  geom_jitter() + 
  geom_point(data=Orig_agg, cex=3) + 
  geom_line(data=Orig_agg,lwd=1) + 
  theme(legend.position="right") +
  ggtitle("#19 Rating for each class offered 2007-2017 w Means for each AY. (Jittered)")