a visualization that draws a correlation network from pairwise correlations between features (dataset columns). this calculation of correlations between each column in a dataset is one method to prepare a graph of relationships from tabular data.
the correlations are calculated by h2o for the famous airlines_all dataset as aggregated by h2o-3. comparisons between numeric columns use Pearson Correlation while comparisons between pairs of two categorical columns as well as a mixed pair of one categorical column and one numeric column use a novel method developed by Lee Wilkinson
this example draws from a static graph.json file produced from the tabular airlines dataset by the h2o server.
community detection with jLouvain
repo for the correlation-graph visualization component shown in this example: https://github.com/micahstubbs/correlation-graph