A study looking at data judgement accuracy in pie charts with shape variations.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
analysis
public
README.md
app.js
package.json
redis.conf

README.md

Pie and Donut Chart Evaluation

This project was used in a study that used the Mechanical Turk platform to test a series of pie chart variations.

The study is built on the excellent Experimentr.js project.

Data

The collected data files are in the analysis directory, together with some scripts and the R code to create the figures in the paper.

Data Files

  • pie-variations-data.json: the original data as collected during the study, in JSON format
  • pie-variations-reshaped-unique.csv: trial data in CSV format, with incomplete trials removed
  • pie-variations-demographics.csv: demographics data in CSV format
  • predictions.csv: predicted values for the different charts based on area and arc length, for all angles in half-degree steps
  • pie-variations-enriched.csv: same as pie-variations-reshaped-unique.csv but with additional columns for predicted values based on arc length and area

Scripts

  • enrich.py: creates the predictions.csv file and adds predictions to the trials data to create pie-variations-enriched.csv
  • cleanup.py: parses the original JSON data and reshapes the data into one row per trial, generates pie-variations-reshaped-unique.csv and pie-variations-demographics.csv
  • variationviolins.R: R code to do analysis and generate the figures in the paper

Running The Study

To run the project, you'll need Redis and Node.js.

To start the Redis server, run the following command from the project directory:

redis-server redis.conf

The Node server works on port 80, so it needs root access, and the project is set to use Forever.js to ensure it keeps running:

forever start app.js

Once you have the project running, you can visit localhost in your browser to see the survey.