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$timestamp)
  11
  12 ## drop calls where the person wasn't present
  13 call.list[!call.list$answered,]
  14 call.list <- call.list[call.list$answered,]
  15
  16 call.counts <- data.frame(table(call.list$discord.name))
  17 colnames(call.counts) <- c("discord.name", "num.calls")
  18
  19 d <- merge(d, call.counts, all.x=TRUE, all.y=TRUE, by="discord.name"); d
  20
  21 ## set anything that's missing to zero
  22 d$num.calls[is.na(d$num.calls)] <- 0
  23
  24 attendance <- unlist(lapply(list.files(".", pattern="^attendance-.*tsv$"), function (x) {d <- read.delim(x); strsplit(d[[2]], ",")}))
  25
  26
  27 file.to.attendance.list <- function (x) {
  28     tmp <- read.delim(x)
  29     d.out <- data.frame(discord.name=unlist(strsplit(tmp[[2]], ",")))
  30     d.out$day <- rep(as.Date(tmp[[1]][1]), nrow(d.out))
  31     return(d.out)
  32 }
  33
  34 attendance <- do.call("rbind",
  35                       lapply(list.files(".", pattern="^attendance-.*tsv$"),
  36                              file.to.attendance.list))
  37
  38 attendance.counts <- data.frame(table(attendance$discord.name))
  39 colnames(attendance.counts) <- c("discord.name", "num.present")
  40
  41 d <- merge(d, attendance.counts,
  42            all.x=TRUE, all.y=TRUE,
  43            by="discord.name")
  44
  45 days.list <- lapply(unique(attendance$day), function (day) {
  46     day.total <- table(call.list$day == day)[["TRUE"]]
  47     lapply(d$discord.name, function (discord.name) {
  48         num.present <- nrow(attendance[attendance$day == day & attendance$discord.name == discord.name,])
  49         data.frame(discord.name=discord.name,
  50                    days.present=(num.present/day.total))
  51     })
  52 })
  53
  54 days.tmp <- do.call("rbind", lapply(days.list, function (x) do.call("rbind", x)))
  55
  56 days.tbl <- tapply(days.tmp$days.present, days.tmp$discord.name, sum)
  57
  58 attendance.days <- data.frame(discord.name=names(days.tbl),
  59                               days.present=days.tbl,
  60                               days.absent=length(list.files(".", pattern="^attendance-.*tsv$"))-days.tbl)
  61
  62 d <- merge(d, attendance.days,
  63            all.x=TRUE, all.y=TRUE, by="discord.name")
  64 d[sort.list(d$days.absent), c("discord.name", "num.calls", "days.absent")]
  65
  66 ## make some visualizations of whose here/not here
  67 #######################################################
  68
  69 png("questions_absence_histogram_combined.png", units="px", width=800, height=600)
  70
  71 ggplot(d) +
  72     aes(x=as.factor(num.calls), fill=days.absent, group=days.absent) +
  73     geom_bar(color="black") +
  74     scale_x_discrete("Number of questions asked") +
  75     scale_y_continuous("Number of students") +
  76     scale_fill_continuous("Days absent", low="red", high="blue")+
  77     theme_bw()
  78
  79 dev.off()
  80
  81 png("questions_absenses_boxplots.png", units="px", width=800, height=600)
  82
  83 ggplot(data=d) +
  84     aes(x=as.factor(num.calls), y=days.absent) +
  85     geom_boxplot() +
  86     scale_x_discrete("Number of questions asked") +
  87     scale_y_continuous("Days absent")
  88
  89 dev.off()
  90