aariq / crayonbracket

Analysis of #crayonbracket

#crayonbracket is a bracket style tournament on Twitter organized by @sbarolo. I used the rtweet and httr R packages to access the Twitter poll results for each match-up and pulled the vote totals for each color as well as the hex codes for the color. I then used the colorspace R package to convert the hex codes into the LUV color space.

Code in get_tweet_data.R accesses the Twitter API to get the data (you'll need a bearer token for this to run on your machine).

Code in color_distance.R takes that data, does some color space transformations, and makes a plot.

Cleaned data is provided in the data/ folder

Does color difference affect win margin?

In other words, if two colors are very perceptually different, does the winning color win by more votes?

(more to come...)

Contributing

If you know more about human color perception or crayons than I do and want to play around with these data, please feel free to make a pull request, open an issue, or tweet at me.

crayonbracket's Issues

Modeling

Win margin vs. color distance:

• account for number of votes---a margin of 20 is big for 60 total votes but small for 1000 votes.
• Poisson with an offset?
• Add best fit line to plot

Votes vs. saturation, hue, etc.

• Figure out best colorspace for this analysis
• Include 3 dimensions as predictors
• Also account for total votes with an offset
• determine if one or more dimensions best determines number of votes / total votes