This is a simple R package for working with FaceReader 6.xx output data because the default output is relatively difficult to wrangle and utilize in any statistics program without major manual overhaul. This package helps streamline the process with a few additional functions I find useful.

Note: This package is currently under development and not a complete package by any means.

You can install the development version of the facereadeR package using devtools:

install.packages('devtools')
devtools::install_github('d-bohn/facereadeR')
library(facereadeR)

After installing the package, you can start to analyze FaceReader output with just a few simple function calls that read in and format the data, clean it, and generally tidy it up for further statistical analyses. To get started, try reading in the included example data files using:

files <- list.files(system.file("extdata/", package="facereadeR"), full.names = TRUE)
library(purrr);library(dplyr)

data <- files %>% purrr::map(., read_facereader, as.is = TRUE) %>% 
  reduce(., rbind)

This data set includes raw FaceReader output (the text file that FaceReader 6.1 produces) from a selection of the first 2016 Republican debate where reporter Megyn Kelly confronts Presidential candidate Donald Trump on recent sexist remarks. There are two files: one for Kelly where she asks the question, and one for Trump where he responds.

After the data are imported into a dataframe, we typically need to remove failures to find the face and replace them with NA values. Additionally, we will create appropriate factors for each of the variables to be analyzed.

data <- clean_data(data, include = 'Basic')

If we are interested in analyzing Action Units (AUs), and they are provided in the output, we could run the following to rename them to a simpler format as well as score them so that they become numeric values instead of A, B, C, D.

data <- rename_aus(data)
data <- score_aus(data)

At this point, the data is in a tidy format having been read in and aggregated together. Additionally, the AUs have been converted to numeric values for quantitative analysis.

head(data[1:10])