Section 3 Data
The main source of data used in this report is the download logs from the RStudio CRAN mirror site : https://cran.rstudio.com/. These log files are created for every instance of download of an R-package via the RStudio CRAN mirror, then these log files are processed, daily, into CSV files that contain the following variables with the name of header in brackets:
- Date (
date), - Time in UTC time zone (
time), - Size of the file in bytes (
size), - Version of R used to download the package (
r_version), - Architecture type for R (i386 = 32 bit, x86_64 = 64 bit) (
r_arch), - Operating System (darwin9.8.0 = mac, mingw32 = windows) (
r_os), - Package (
package), - Country in two letter ISO country code (
country), and - Anonymised daily unique id (
ip_id).
A similar log file is also created for every download of R from the RStudio CRAN mirror with the processed log file generating a CSV file that contains the same variables except r_arch and package, and r_version and r_os are named as version and os. These CSV files are hosted at http://cran-logs.rstudio.com/ and updated daily with data available from 1st October 2012.
The log files of a particular day is processed and compressed into a single CSV file of about 40 megabytes (file sizes of earlier years are much smaller due to lower number of download logs). As there are over 700,000 CSV files, a simple estimate of the size of the data is 28 terabytes - far exceeding typical portable hard drives which are 1-4 terabytes.
The summarised version of data, where the data show the total daily download counts for each package, is accessible using the cranlogs R-package. The cranlogs package accesses this summary data through the web application programming interface (API) maintained by r-hub (R-Hub 2020).