data/track_participation.R

   1 library(ggplot2)
   2
   3 gs <- read.delim("student_information.tsv")
   4 d <- gs[,c(2,5)]
   5 colnames(d) <- c("student.num", "discord.name")
   6
   7 call.list <- do.call("rbind", lapply(list.files(".", pattern="^call_list-.*tsv$"), function (x) {read.delim(x)[,1:3]}))
   8 colnames(call.list) <- gsub("_", ".", colnames(call.list))
   9
  10 call.list$day <- as.Date(call.list$timestamp)
  11
  12 call.counts <- data.frame(table(call.list$discord.name))
  13 colnames(call.counts) <- c("discord.name", "num.calls")
  14
  15 d <- merge(d, call.counts, all.x=TRUE, all.y=TRUE, by="discord.name"); d
  16
  17 ## set anything that's missing to zero
  18 d$num.calls[is.na(d$num.calls)] <- 0
  19
  20 attendance <- unlist(lapply(list.files(".", pattern="^attendance-.*tsv$"), function (x) {d <- read.delim(x); strsplit(d[[2]], ",")}))
  21
  22
  23 file.to.attendance.list <- function (x) {
  24     tmp <- read.delim(x)
  25     d.out <- data.frame(discord.name=unlist(strsplit(tmp[[2]], ",")))
  26     d.out$day <- rep(as.Date(tmp[[1]][1]), nrow(d.out))
  27     return(d.out)
  28 }
  29
  30 attendance <- do.call("rbind",
  31                       lapply(list.files(".", pattern="^attendance-.*tsv$"),
  32                              file.to.attendance.list))
  33
  34 attendance.counts <- data.frame(table(attendance$discord.name))
  35 colnames(attendance.counts) <- c("discord.name", "num.present")
  36
  37 d <- merge(d, attendance.counts,
  38            all.x=TRUE, all.y=TRUE,
  39            by="discord.name")
  40
  41 days.list <- lapply(unique(attendance$day), function (day) {
  42     day.total <- table(call.list$day == day)[["TRUE"]]
  43     lapply(d$discord.name, function (discord.name) {
  44         num.present <- nrow(attendance[attendance$day == day & attendance$discord.name == discord.name,])
  45         data.frame(discord.name=discord.name,
  46                    days.present=(num.present/day.total))
  47     })
  48 })
  49
  50 days.tmp <- do.call("rbind", lapply(days.list, function (x) do.call("rbind", x)))
  51
  52 days.tbl <- tapply(days.tmp$days.present, days.tmp$discord.name, sum)
  53
  54 attendance.days <- data.frame(discord.name=names(days.tbl),
  55                               days.present=days.tbl,
  56                               days.absent=length(list.files(".", pattern="^attendance-.*tsv$"))-days.tbl)
  57
  58 d <- merge(d, attendance.days,
  59            all.x=TRUE, all.y=TRUE, by="discord.name")
  60 d[sort.list(d$days.absent), c("discord.name", "num.calls", "days.absent")]
  61
  62 ## make some visualizations of whose here/not here
  63 #######################################################
  64
  65 png("questions_absence_histogram_combined.png", units="px", width=800, height=600)
  66
  67 ggplot(d) +
  68     aes(x=as.factor(num.calls), fill=days.absent, group=days.absent) +
  69     geom_bar(color="black") +
  70     scale_x_discrete("Number of questions asked") +
  71     scale_y_continuous("Number of students") +
  72     scale_fill_continuous("Days absent", low="red", high="blue")+
  73     theme_bw()
  74
  75 dev.off()
  76
  77 png("questions_absenses_boxplots.png", units="px", width=800, height=600)
  78
  79 ggplot(data=d) +
  80     aes(x=as.factor(num.calls), y=days.absent) +
  81     geom_boxplot() +
  82     scale_x_discrete("Number of questions asked") +
  83     scale_y_continuous("Days absent")
  84
  85 dev.off()
  86