从数据框中提取日期并使用R绘制时间序列

Question

I would like to plot a time series which presents the number of log s hourly. I tried at first to split the the date of every log from the dataframe in order to count the number of log s hourly.

I have the following dataframe :

[Fri Jun  1 15:56:37 1995] httpd: send aborted for disarray.demon.co.uk
[Fri Jun  1 16:29:29 1995] httpd: send aborted for ansc86024.usask.ca
[Fri Jun  1 16:31:42 1995] httpd: send aborted for 194.20.24.70
[Fri Jun  1 16:34:11 1995] httpd: send aborted for sw24-70.iol.it
[Fri Jun  1 16:41:02 1995] httpd: send aborted for educ026.usask.ca
[Fri Jun  1 16:41:13 1995] httpd: send aborted for educ026.usask.ca
[Fri Jun  1 16:41:13 1995] httpd: send aborted for sw24-70.iol.it
[Fri Jun  1 16:45:07 1995] httpd: send aborted for 128.233.18.38
[Fri Jun  1 17:26:50 1995] httpd: send aborted for pc117c.nwrel.org
[Fri Jun  1 17:46:53 1995] httpd: send aborted for geoff.usask.ca
[Fri Jun  2 17:57:09 1995] httpd: send aborted for piweba3y.prodigy.com
[Fri Jun  2 17:57:50 1995] httpd: send aborted for piweba3y.prodigy.com
[Fri Jun  2 18:10:15 1995] httpd: send aborted for 193.74.92.109
[Fri Jun  2 20:14:30 1995] httpd: send aborted for 128.233.13.41
[Fri Jun  2 20:15:59 1995] httpd: send aborted for peter.net4.io.org
[Fri Jun  2 21:11:54 1995] httpd: send aborted for ped374.usask.ca

I want to get the following plot with the number of log s hourly :

I tried to add the date column using the gsub function :

df$date <- gsub(".+[(.*)]","",df[0])

Answer 1

How about this:

# Data in form of a string vector
dat = c("[Fri Jun 1 15:56:37 1995] httpd: send aborted for disarray.demon.co.uk", 
        "[Fri Jun 1 16:29:29 1995] httpd: send aborted for ansc86024.usask.ca", 
        "[Fri Jun 1 16:31:42 1995] httpd: send aborted for 194.20.24.70", 
        "[Fri Jun 1 16:34:11 1995] httpd: send aborted for sw24-70.iol.it", 
        "[Fri Jun 1 16:41:02 1995] httpd: send aborted for educ026.usask.ca", 
        "[Fri Jun 1 16:41:13 1995] httpd: send aborted for educ026.usask.ca", 
        "[Fri Jun 1 16:41:13 1995] httpd: send aborted for sw24-70.iol.it", 
        "[Fri Jun 1 16:45:07 1995] httpd: send aborted for 128.233.18.38", 
        "[Fri Jun 1 17:26:50 1995] httpd: send aborted for pc117c.nwrel.org", 
        "[Fri Jun 1 17:46:53 1995] httpd: send aborted for geoff.usask.ca", 
        "[Fri Jun 2 17:57:09 1995] httpd: send aborted for piweba3y.prodigy.com", 
        "[Fri Jun 2 17:57:50 1995] httpd: send aborted for piweba3y.prodigy.com", 
        "[Fri Jun 2 18:10:15 1995] httpd: send aborted for 193.74.92.109", 
        "[Fri Jun 2 20:14:30 1995] httpd: send aborted for 128.233.13.41", 
        "[Fri Jun 2 20:15:59 1995] httpd: send aborted for peter.net4.io.org", 
        "[Fri Jun 2 21:11:54 1995] httpd: send aborted for ped374.usask.ca")

library(dplyr)
library(lubridate)

Extract date string:

dat = data.frame(date.string = gsub(".{5}(.*)\\].*", "\\1", dat))

Convert date string to POSIXct datetime format:

dat$date = as.POSIXct(dat$date.string, format= "%b %e %H:%M:%S %Y")

Now, summarise by hour. We throw away the minutes and seconds so that we can then just group by date to get counts by hour:

datByHour = dat %>% 
  mutate(date = as.POSIXct(paste0(paste(year(date),month(date),day(date),sep="-"), 
                                  " ", 
                                  paste(hour(date),"00:00", sep=":")))) %>%
  group_by(date) %>%
  tally 

datByHour

  date n 1 1995-06-01 15:00:00 1 2 1995-06-01 16:00:00 7 3 1995-06-01 17:00:00 2 4 1995-06-02 17:00:00 2 5 1995-06-02 18:00:00 1 6 1995-06-02 20:00:00 2 7 1995-06-02 21:00:00 1 

Plot hourly counts:

ggplot(datByHour, aes(date, n)) + 
  geom_line(aes(group=1)) +
  scale_x_datetime(date_labels="%b %e, %Y: %H")