cran.stats

cran.stats is a package that provides functions to explore and visualise hourly, daily, weekly, monthly or yearly package download stats from Rstudio cran mirror logs. It also implements an approach to calculate the number of downloads for a given package alone by identifying and subtracting the downloads due to it's dependent packages.

Installation

To install the package, use devtools:

devtools::install_github("arunsrinivasan/cran.stats")

Usage

There are three functions, to be used in the order shown below:

read_logs

read_logs(start, end, path="./", dir="cran-mirror", verbose=TRUE)

The function takes a start and end date and downloads all the logs available from the Rstudio mirror for that duration and places them in the directory path/dir. The downloaded files are automatically replaced with their unzipped versions. If an unzipped log file already exists, then the downloads are skipped. And if the downloaded file was corrupted (which would error while unzipping), then we attempt to download it just once again. If it fails again, we skip this log.

As an example, let's download all the logs between July 01, 2014 and July 31, 2014.

dt = read_logs(start = as.Date("2014-07-01"), 
               end = as.Date("2014-07-31"), 
               dir="cran-mirror", 
               verbose = TRUE)

dim(dt)
# [1] 4824243       5

key(dt)
# [1] "package" "date"    "time"   

dt["data.table"]
#          package       date     time country ip_id
#     1: data.table 2014-07-01 00:03:47      US    13
#     2: data.table 2014-07-01 00:04:30      US    13
#     3: data.table 2014-07-01 00:07:12      US     2
#     4: data.table 2014-07-01 00:09:10      US    51
#     5: data.table 2014-07-01 00:12:33      US    13
#    ---                                             
# 14393: data.table 2014-07-31 23:33:04      US  7159
# 14394: data.table 2014-07-31 23:33:09      US  7100
# 14395: data.table 2014-07-31 23:35:16      NZ  1185
# 14396: data.table 2014-07-31 23:52:41      US 11181
# 14397: data.table 2014-07-31 23:57:21      US  7159

stats_logs

stats_logs(dt, type="monthly", packages=c("data.table"), 
          dependency=TRUE, duration = 30L)

Have a look at ?stats_logs for more info. Let's get the daily download stats for the month of July.

dt_s = stats_logs(dt, type="daily")
#        package        key tot_N dep_N
#  1: data.table 2014-07-01   456    94
#  2: data.table 2014-07-02   406    75
#  3: data.table 2014-07-03   393    59
#  4: data.table 2014-07-04   267    50
#  5: data.table 2014-07-05   178    46
#  6: data.table 2014-07-06   169    32
#  7: data.table 2014-07-07   450    73
#  8: data.table 2014-07-08   680   120
#  9: data.table 2014-07-09   663    97
# 10: data.table 2014-07-10   705    82
# 11: data.table 2014-07-11   709   116
# 12: data.table 2014-07-12   556    44
# 13: data.table 2014-07-13   676    28
# 14: data.table 2014-07-14   605    77
# 15: data.table 2014-07-15   637    67
# 16: data.table 2014-07-16   549    92
# 17: data.table 2014-07-17   586    68
# 18: data.table 2014-07-18   481    74
# 19: data.table 2014-07-19   230    42
# 20: data.table 2014-07-20   273    38
# 21: data.table 2014-07-21   525   105
# 22: data.table 2014-07-22   504    67
# 23: data.table 2014-07-23   605    90
# 24: data.table 2014-07-24   530    89
# 25: data.table 2014-07-25   394    70
# 26: data.table 2014-07-26   214    26
# 27: data.table 2014-07-27   215    32
# 28: data.table 2014-07-28   440   108
# 29: data.table 2014-07-29   456    87
# 30: data.table 2014-07-30   440    81
# 31: data.table 2014-07-31   405    71
#        package        key tot_N dep_N

plot_logs

plot_logs(dt)

Let's now plot it. Note that it detects the presence of dep_N column and automatically draws two plots, first with just the direct downloads for the packages specified, and the next with the number of downloads just due to dependencies.

plot_logs(dt_s)