CREATED BY: Nicolas Fernandez
We visualized the high-dimensional member data using our in-development interactive WebGL heatmap Jupyter widget called Clustergrammer2. Users are visualized as columns, country/city as column categories, and dimensions are visualized as rows. The interactive heatmap allows users to interactively explore all 3,500 member's data, reorder members based on attributes (e.g. sum/variance), and identify clusters of members based on similar attributes. We also calculated country "signatures" based on the average properties of member data in the country.
PROCESS FOR CREATION:
We looked up country and city information using geopy and added this information as user categories (column categories). Member dimensions (e.g. society score) were Z-scored across members to help compare different measurement types. The notebook can be viewed on NBViewer (https://nbviewer.jupyter.org/github/cornhundred/datavizsociety/blob/master/notebooks/3.0_Viz_Heatmaps.ipynb), run on mybinder (https://mybinder.org/badge_logo.svg)](https://mybinder.org/v2/gh/cornhundred/datavizsociety/master?filepath=notebooks%2F3.0_Viz_Heatmaps.ipynb), the code is on GitHub (https://github.com/cornhundred/datavizsociety), and a quick demo video can be seen on twitter (https://twitter.com/franschrandez/status/1105529159604989953)