RSiteCatalyst Version 1.2 Release Notes

Version 1.2 of the RSiteCatalyst package to access the Adobe Analytics API is now available on CRAN! Changes include:

  • Removed RCurl package dependency
  • Changed argument order for GetAdminConsoleLog to avoid error when date not passed
  • Return proper numeric type for metric columns
  • Fixed bug in GetEVars function
  • Added validate:true flag to API to improve error reporting
  • Removed remaining references to Omniture

For the most part, the only noticeable change for most users will be that you no longer need to call as.numeric() on a DataFrame after getting the results of an API call, as all functions now return the proper numeric type.

Changes from Development Version

For any of you out there that may have installed the 1.2 development version directly from GitHub, the only difference between the 1.2 development version and the stable, CRAN version of the package is that support for the Adobe Analytics Real Time API has been removed. This functionality will continue to be developed on the 1.3 development branch on GitHub.


Testing

For this release, I’ve made a more concerted effort to test RSiteCatalyst on various platforms outside of OSX (where I do my development). RSiteCatalyst works in the following environments:

  • OSX Lion and prior
  • Ubuntu 12.04 LTS
  • Windows 7 64-bit SP1
  • Windows 8.1 64-bit
  • R 2.15.2 and newer
  • R and RStudio

If your environment is not listed above, it is still likely the case that RSiteCatalyst will work in your environment, as there is no operating-system-specific code in the package. If you are finding issues, validate that you have all package dependencies installed, your Adobe account has Web Service Access privileges (set in Admin panel), you have permission access to the report suites you are trying to access (also an Admin panel setting) and that your company doesn’t have any firewall settings that would prevent API access.

Support

If you run into any problems with RSiteCatalyst, please file an issue on GitHub so it can be tracked properly. Note that I’m not an Adobe employee, so I can only provide so much support, as in most cases I can’t validate your settings to ensure you are set up correctly (nor do I have any inside information about how the system works :) )

 

Comments

  1. Hey Randy,

    thanks for this awesome package. It really helps automate a ton of reports and makes the analysis a lot more efficient.

    In order for us to extract bigger amounts of data we had to tweak the package a little. We extended the While loop up to 500 iterations. We were wondering if you had an answer why there are so many differences in the number of attempts used to get reports from the data warehouse.
    Is this the same as the loop the report builder seems to be stuck in from time to time?
    Do you know of a way to speed the data extraction up?

    Greetings from Germany
    Daniel & Max

  2. Randy Zwitch says:

    Thanks for stopping by Max/Daniel, glad you are finding the package useful!

    I’m not an Adobe employee, but as far as I know Report Builder uses the API, so they are in fact the same source (and the same delays would apply). I believe the reason why there is a delay is that the API is calculating the information on the fly, when you use two elements it actually needs to run the query against a MySQL database (I think). So it can often be faster to submit a series of smaller calls, such as the same report with 1 day granularity 30 times than to request one month at a time.

    I am aware that the looping timeout isn’t optimal, and I’ve fixed that for version 1.3, I just haven’t had time to submit that to CRAN yet:

    https://github.com/randyzwitch/RSiteCatalyst/issues/23

    If you’re okay with using “beta” software, you can download v1.3 using devtools by Hadley Wickham:

    install.packages(“devtools”)
    library(devtools)

    install_github(“RSiteCatalyst”, “randyzwitch”, ref = “version_1_3″)
    library(RSiteCatalyst)

    • Hi Randy,

      thanks for the answer. I just saw it today. I was waiting for an mail alert if a comment was postet :). We changed the loop to call for small pieces of data and it really works a lot better.

      Thanks for the link to the beta version. I’ll try it out.

      Greetings
      Max

Leave a Reply

%d bloggers like this: