The original data is in the .xlsx file. It's a spreadsheet of ~730 rocks with their associated minerals.
I deleted the first three columns and saved it as rocks.csv
, then imported the .csv into R (header = TRUE)
.
color.csv
is the groupings which the domain scientist assigned to the 51 minerals. Each group is assign a specific color in the visualizations.
How distances are calculated on the MDS maps:
mineral 1 is contained in (a + b) types of rocks
mineral 2 is contained in (b + c) types of rocks
(b types of rocks contain both mineral 1 and mineral 2)
distance between mineral 1 and mineral 2 is calculated as: 1 - b / min((a + b), (b + c))
on macOS, XQuartz needs to be installed along with the rgl R package to make the 3d plots work.
@misc{morrison2017network,
title={Network analysis of mineralogical systems},
author={Morrison, Shaunna M and Liu, Chao and Eleish, Ahmed and Prabhu, Anirudh and Li, Congrui and Ralph, Jolyon and Downs, Robert T and Golden, Joshua J and Fox, Peter and Hummer, Daniel R and others},
year={2017},
publisher={Mineralogical Society of America}
}