setwd("/Users/msonnabaum")

require(ggplot2)
require(lubridate)
require(tools)
require(stringr)


df <- read.csv("wmg_response_times2.csv")
df$datetime <- ymd_hms(as.character(df$time))

df$ext <- as.factor(file_ext(str_extract(basename(as.character(df$url)), "[^?]*")))

quantiles <- c(0, 0.25, 0.5, 0.75, 0.95, 0.95, 1)

quantile(df[df$host == "slamonitor.wmg-gardens.com", ]$response_time, quantiles)

##   0%  25%  50%  75%  95%  95% 100% 
##   80  469  611  885 1765 1765 5371

quantile(df[df$host == "tangle002.wmg-egardens.acquia-sites.com", ]$response_time, 
    quantiles)

##      0%     25%     50%     75%     95%     95%    100% 
##    98.0   210.0   366.5   933.0  6472.4  6472.4 42488.0

ggplot(df, aes(ext, response_time, fill = host)) + geom_boxplot(aes(ext))