README.md

In burchill/pupilr: Manage and Analyze The Pupillometry Project

pupilr: a package to manage a series of eye-tracking experiments

This package is the public version of the one which manages a pupillometry (eye-tracking) project. (The original R package is stored as part of a private repository on Bitbucket for confidentiality reasons.) This modified package is put online to showcase the development work I've done on it.

The original package has three primary uses:

• First, it pre-processes the raw data (which has personally identifiable information and is thus stored on a secure remote server) when run on the data server so that the accessible data no longer has participant information in it.

• Second, it bundles this data together in a nice, manageable package (for me).

• Third, it provides a number of helpful functions to further process the eye-tracking data for analysis and plotting.

Seeing it in action

I've included the PDFs and .Rmd files of two internal reports I've made using this package in the vignettes folder. You can use these to get a sense of how the code works in action, as well as how I code. Unfortunately, these aren't "real" vignettes in the sense that they are unbuildable in this package (see below).

Caveats:

Although this package was designed with documentation and other user/developer-friendly aspects, there are a few essential things that prevent it from being CRAN-worthy, or even super distributable.

Most importantly, I did NOT include Imports or Suggests in the DESCRIPTION file. I do have parts of the code that check to make sure certain packages are installed, but I had a traumatic incident with making my own R package once have been mentally scarred a little. This code is actually (sadly) based on dplyr 0.5, and not tested on the new versions. I admit all that is dumb, but this package is primarily for my own uses, so I'll get around to fixing all that stuff later.

Also, this package does NOT include the actual participant data. Although the data processing flow of this package removes the most personally identifiable participant information (i.e., participants' Amazon Mechanical Turker IDs), certain demographic information about participants is retained. This is ok if the data on my lab computer, but I want to be sure that I don't cross any IRB lines by publicly sharing that data. You can get a sense of what the data would look like, and how it would be processed with the script data-raw/data_preprocessing_cleanTurkData.R.

This also means that the "vignettes" that are included in this package are not rebuildable. This would be true anyway, as the "vignettes" are actually just preliminary reports I did for these experiments. You'll just have to be content with seeing what they would look like had you had access to the participant data.

Finally, although I have a detailed git history of the development of this package on the private repository on Bitbucket, importing that could reveal personally identifiable participant information, so this git history is starting in media res.

Acknowledgments

I have to acknowledge Dave Kleinschmidt for his advice on managing data in R packages. His tutorials were very helpful in establishing a clean workflow.

I also have to acknowledge Kodi Weatherholtz for the work he did on a related project. The flow of the vignettes here, and some of the visualization techniques, are based on his amazing data viz skills.

burchill/pupilr documentation built on May 22, 2019, 2:27 p.m.