commit a1a80237e29fb46101ecfdf126ecbe6f02e0ade1 Author: Karsten Loesing karsten.loesing@gmx.net Date: Wed Jul 25 17:09:00 2012 +0200

Update #6443 graphing code. --- task-6443/.gitignore | 4 +++ task-6443/cumulated-weights.R | 47 ++++++++++++++++++++++++++++++++++------- 2 files changed, 43 insertions(+), 8 deletions(-)

diff --git a/task-6443/.gitignore b/task-6443/.gitignore new file mode 100644 index 0000000..6a3a120 --- /dev/null +++ b/task-6443/.gitignore @@ -0,0 +1,4 @@ +*.csv +*.pdf +*.png + diff --git a/task-6443/cumulated-weights.R b/task-6443/cumulated-weights.R index 50d0f4c..c324acb 100755 --- a/task-6443/cumulated-weights.R +++ b/task-6443/cumulated-weights.R @@ -2,10 +2,40 @@ require(ggplot2) require(scales) require(reshape)

-c <- read.csv("cumulated-weights.csv") -c <- c[c$top_relays %in% c(1, 2, 3, 5, 10, 15, 20, 30, 50), ] +cw <- read.csv("cumulated-weights.csv", stringsAsFactors = FALSE) +v <- sort(as.POSIXlt(unique(cw$validafter), tz = "UTC"), + decreasing = TRUE) +now <- week <- month <- threemonths <- year <- v[1] +week$mday <- week$mday - 7 +month$mon <- month$mon - 1 +threemonths$mon <- threemonths$mon - 3 +year$year <- year$year - 1 +v <- c( + format((v[v <= year])[1], usetz = FALSE), + format((v[v <= threemonths])[1], usetz = FALSE), + format((v[v <= month])[1], usetz = FALSE), + format((v[v <= week])[1], usetz = FALSE), + format(now, usetz = FALSE)) +v <- cw[cw$validafter %in% v, ] +v <- data.frame(x = v$top_relays, y = v$total_exit_probability, + colour = factor(v$validafter, levels = rev(levels(factor(v$validafter))))) +ggplot(v, aes(x = x, y = y, colour = colour)) + +geom_point() + +geom_line() + +scale_x_continuous(name = "\nTop-x relays by exit probability") + +scale_y_continuous(name = "Total exit probability\n", labels = percent) + +scale_colour_hue(name = "", + labels = c(format(as.Date(now), format = "%B %d, %Y"), "1 week before", + "1 month before", "3 months before", "1 year before")) + +opts(title = paste("Probability of selecting one of the top-x relays for", + "the exit position\n")) +ggsave("exit-probability-cdf-a.png", width = 8, height = 5, dpi = 100) + +c <- cw +c <- c[c$top_relays %in% c(1, 2, 5, 10, 20, 50), ] c <- aggregate(list(total_exit_probability = c$total_exit_probability), - by = list(date = as.Date(c$validafter, origin = "1970-01-01 00:00:00"), + by = list(date = as.Date(cut.Date(as.Date(c$validafter, + origin = "1970-01-01 00:00:00"), "week")), top_relays = c$top_relays), FUN = mean) ggplot(c, aes(x = date, y = total_exit_probability, colour = as.factor(top_relays))) + @@ -16,12 +46,13 @@ scale_y_continuous(name = "Total exit probability of top-x relays\n", scale_colour_hue(name = "Top-x relays by exit probability") + opts(title = paste("Probability of selecting one of the top-x relays for", "the exit position\n"), legend.position = "bottom") -ggsave("exit-probability-cdf.png", width = 8, height = 5, dpi = 100) +ggsave("exit-probability-cdf-b.png", width = 8, height = 5, dpi = 100)

-i <- read.csv("inverse-cumulated-weights.csv") -i <- i[i$total_exit_probability > 0.1 & i$total_exit_probability < 0.9, ] +i <- read.csv("inverse-cumulated-weights.csv", stringsAsFactors = FALSE) +i <- i[i$total_exit_probability %in% factor(c(0.3, 0.4, 0.5, 0.6, 0.7)), ] i <- aggregate(list(top_relays = i$top_relays), - by = list(date = as.Date(i$validafter, origin = "1970-01-01 00:00:00"), + by = list(date = as.Date(cut.Date(as.Date(i$validafter, + origin = "1970-01-01 00:00:00"), "week")), total_exit_probability = i$total_exit_probability), FUN = mean) ggplot(i, aes(x = date, y = top_relays, colour = paste(100 * total_exit_probability, "%", sep = ""))) + @@ -31,5 +62,5 @@ scale_y_continuous(name = "Top-x relays by exit probability\n") + scale_colour_hue(name = "Total exit probability of top-x relays") + opts(title = paste("Number of relays making up the top-x for a given", "total exit probability\n"), legend.position = "bottom") -ggsave("exit-probability-icdf.png", width = 8, height = 5, dpi = 100) +ggsave("exit-probability-cdf-c.png", width = 8, height = 5, dpi = 100)