assessment_and_tracking/track_participation.R

   1 setwd("~/online_communities/coldcallbot/data/")
   2
   3 library(ggplot2)
   4 library(data.table)
   5
   6 gs <- read.delim("student_information.tsv")
   7 d <- gs[,c(2,4)]
   8 colnames(d) <- c("student.num", "teams.name")
   9
  10 call.list <- do.call("rbind", lapply(list.files(".", pattern="^call_list-.*tsv$"), function (x) {read.delim(x, stringsAsFactors=FALSE)[,1:4]}))
  11
  12 colnames(call.list) <- gsub("_", ".", colnames(call.list))
  13
  14 table(call.list$unique_name)
  15
  16 call.list$day <- as.Date(call.list$timestamp)
  17
  18 ## drop calls where the person wasn't present
  19 call.list.full <- call.list
  20 call.list[!call.list$answered,]
  21 call.list <- call.list[call.list$answered,]
  22
  23 call.counts <- data.frame(table(call.list$discord.name))
  24 colnames(call.counts) <- c("discord.name", "num.calls")
  25
  26 d <- merge(d, call.counts, all.x=TRUE, all.y=TRUE, by="discord.name"); d
  27
  28 ## set anything that's missing to zero
  29 d$num.calls[is.na(d$num.calls)] <- 0
  30
  31 attendance <- unlist(lapply(list.files(".", pattern="^attendance-.*tsv$"), function (x) {d <- read.delim(x); strsplit(d[[2]], ",")}))
  32
  33 file.to.attendance.list <- function (x) {
  34     tmp <- read.delim(x)
  35     d.out <- data.frame(discord.name=unlist(strsplit(tmp[[2]], ",")))
  36     d.out$day <- rep(as.Date(tmp[[1]][1]), nrow(d.out))
  37     return(d.out)
  38 }
  39
  40 attendance <- do.call("rbind",
  41                       lapply(list.files(".", pattern="^attendance-.*tsv$"),
  42                              file.to.attendance.list))
  43
  44 ## create list of folks who are missing in class
  45 missing.in.class  <- call.list.full[is.na(call.list.full$answered) |
  46                                     (!is.na(call.list.full$answered) & !call.list.full$answered),
  47                                     c("discord.name", "day")]
  48
  49 missing.in.class <- unique(missing.in.class)
  50
  51 setDT(attendance)
  52 setkey(attendance, discord.name, day)
  53 setDT(missing.in.class)
  54 setkey(missing.in.class, discord.name, day)
  55
  56 ## drop presence for people on missing days
  57 attendance[missing.in.class,]
  58 attendance <- as.data.frame(attendance[!missing.in.class,])
  59
  60 attendance.counts <- data.frame(table(attendance$discord.name))
  61 colnames(attendance.counts) <- c("discord.name", "num.present")
  62
  63 d <- merge(d, attendance.counts,
  64            all.x=TRUE, all.y=TRUE,
  65            by="discord.name")
  66
  67 days.list <- lapply(unique(attendance$day), function (day) {
  68     day.total <- table(call.list.full$day == day)[["TRUE"]]
  69     lapply(d$discord.name, function (discord.name) {
  70         num.present <- nrow(attendance[attendance$day == day & attendance$discord.name == discord.name,])
  71         if (num.present/day.total > 1) {print(day)}
  72         data.frame(discord.name=discord.name,
  73                    days.present=(num.present/day.total))
  74     })
  75 })
  76
  77 days.tmp <- do.call("rbind", lapply(days.list, function (x) do.call("rbind", x)))
  78
  79 days.tbl <- tapply(days.tmp$days.present, days.tmp$discord.name, sum)
  80
  81 attendance.days <- data.frame(discord.name=names(days.tbl),
  82                               days.present=days.tbl,
  83                               days.absent=length(list.files(".", pattern="^attendance-.*tsv$"))-days.tbl)
  84
  85 d <- merge(d, attendance.days,
  86            all.x=TRUE, all.y=TRUE, by="discord.name")
  87
  88 d[sort.list(d$days.absent), c("discord.name", "num.calls", "days.absent")]
  89
  90 ## make some visualizations of whose here/not here
  91 #######################################################
  92
  93 png("questions_absence_histogram_combined.png", units="px", width=800, height=600)
  94
  95 ggplot(d) +
  96     aes(x=as.factor(num.calls), fill=days.absent, group=days.absent) +
  97     geom_bar(color="black") +
  98     scale_x_discrete("Number of questions asked") +
  99     scale_y_continuous("Number of students") +
 100     scale_fill_continuous("Days absent", low="red", high="blue")+
 101     theme_bw()
 102
 103 dev.off()
 104
 105 png("questions_absenses_boxplots.png", units="px", width=800, height=600)
 106
 107 ggplot(data=d) +
 108     aes(x=as.factor(num.calls), y=days.absent) +
 109     geom_boxplot() +
 110     scale_x_discrete("Number of questions asked") +
 111     scale_y_continuous("Days absent")
 112
 113 dev.off()
 114