projects.mako.cc - github-barcamp-201407/blob - analysis.R

   1 # load up libraries and data
   2 #######################################################################
   3 library(ggplot2)
   4 library(reshape2)
   5 library(parallel)
   6
   7 d <- read.delim("gitdata_1week.tsv", stringsAsFactors=FALSE)
   8
   9 d$url <- NULL
  10 d$date <- gsub("T", " ", d$date)
  11 d$date <- as.POSIXct(d$date)
  12
  13 # crop the event off the end
  14 d$type <- gsub("Event$", "", d$type)
  15
  16 rev(sort(table(d$type)))
  17 head(rev(sort(table(d$name))))
  18 head(rev(sort(table(d$actor))))
  19
  20 d$min <- cut(d$date, breaks="min")
  21 d$hour <- cut(d$date, breaks="hour")
  22
  23 # generate graphs
  24 ######################################################################
  25
  26 pdf("github_graphs.pdf", width=10, height=7)
  27
  28 # graph all together
  29 grid.tmp <- as.data.frame(table(d$hour))
  30 colnames(grid.tmp) <- c("date", "freq")
  31 grid.tmp$date <- as.POSIXct(grid.tmp$date)
  32
  33 qplot(date, freq, data=grid.tmp, geom="line")
  34
  35 # just types of events
  36 grid.tmp <- melt(lapply(tapply(d$type, d$hour, table), as.list), L2~L1)
  37 colnames(grid.tmp) <- c("value", "event", "date")
  38 grid.tmp$date <- as.POSIXct(grid.tmp$date)
  39
  40 ggplot(data=grid.tmp) +
  41     aes(x=date, y=value, group=event, color=event) + geom_line()
  42
  43 ggplot(data=grid.tmp) +
  44     aes(x=date, y=value, group=event, color=event) + geom_bar(stat="identity")
  45
  46 ggplot(data=grid.tmp) +
  47     aes(x=date, y=value, group=event, color=event, size=value) + geom_line() +
  48     facet_grid(event~., scale="free_y")
  49
  50 # create first differences
  51 build.phase.diagram.dataset <- function (d) {
  52     grid.tmp <- as.data.frame(table(d$hour))
  53     colnames(grid.tmp) <- c("date", "freq")
  54     grid.tmp$date <- as.POSIXct(grid.tmp$date)
  55     grid.tmp$freq2 <- c(NA, grid.tmp$freq[1:(length(grid.tmp$freq)-1)])
  56     grid.tmp$diff <- grid.tmp$freq2 - grid.tmp$freq
  57     grid.tmp <- grid.tmp[2:(dim(grid.tmp)[1]-1),]
  58     grid.tmp$hour.of.week <- as.numeric(as.factor(grid.tmp$date))
  59     grid.tmp$hour <- as.numeric(as.factor(grid.tmp$date)) %% 24
  60     return(grid.tmp)
  61 }
  62
  63 phase.diagram <- function (subset) {
  64     grid.tmp <- build.phase.diagram.dataset(d[d$type == subset,])
  65     ggplot(grid.tmp) + aes(x=diff, y=freq2, size=hour, colour=hour) +
  66         geom_path() + ggtitle(paste("Phase Diagram:", subset))
  67
  68 }
  69
  70 grid.tmp <- build.phase.diagram.dataset(d)
  71 ggplot(grid.tmp) + aes(x=diff, y=freq2, size=hour, colour=hour) +
  72     geom_path() + ggtitle("Phase Diagram: All Activity")
  73
  74 phase.diagram("Push")
  75 phase.diagram("Watch")
  76 phase.diagram("Fork")
  77 phase.diagram("PullRequest")
  78 dev.off()