lprisan / jdc-data-analysis

Data preprocessing, analysis scripts and tools for the Journee des Classes experiment

jdc-data-analysis

Data preprocessing, analysis scripts and tools for the Journee des Classes experiment. The project is organized in several R scripts, that go through the different steps of the data analysis process:

• JDCLogPreprocess.R = Pre-processing of logs: from a raw log file to a cleaner R data frame of moments and manipulable positions
• JDCSurveyPreprocess.R = Preprocessing of raw survey data, in csv form
• JDCMapPreprocess.R = Preprocessing of map performance data, from the human video analysis, in csv form
• JDCEyetrackPreprocess.R = Preprocessing of teacher eyetracker data, in txt/csv form
• JDCExploratory.R = Exploratory analysis: First basic analyses of data, to know what we have

Pre-processing

These files normally have a single user function, that does most of the things provided the folder where the raw data is, e.g.:

preprocessJDCLogs(rootDir, doYAMLConversion=FALSE)

These functions will generate one or more .rda files with the clean/processed data for each student/group. These data can be easily loaded in R using:

var <- get(load('filename.rda'))

Note for the log preprocessing: This function will assume that you have a directory rootDir with subdirectories for each lamp log data, named lamp 1, lamp 2 etc. This rootDir will also contain a csv file with each group's beginning and ending times (see logs/ folder in this project). If the doYAMLConversion parameter is set to TRUE, it will try to convert the original pseudo-YAML log files into JSON, a similar format which is more easily read by R (if the parameter is set to FALSE, that means you already did that conversion, and you have .json log files in the lamp folders instead).

Another Note for log preprocessing: The pseudo-YAML files have to be syntactically correct (all braces and parenthesis must be closed, no "half objects" are allowed -- something that happens sometimes in the log file). This is something you will have to check manually

Note on log data: S4G2.rda denotes session 4, Group/Lamp 2. Each row has a numeric timestamp (milliseconds) and 64 more variables with the quadrant in which the element was placed (as character variables with values 0 -- not present, Q1, Q2, Q3, Q4).The quadrants are defined as in classical geometry

Q2	Q1
Q3	Q4

Exploratory analysis

These functions generate some basic descriptive graphics about the data, just run the function:

JDCExplorations(rootDir=".")

This function will take the pre-processed data .rda files (each kind in a different folder, e.g., quests, logs, maps...) and generate a number of graphic files.