Feature tab should auto-summarise distinguishing features
If a feature (e.g. literacy %) is very high in just one cluster, annotate this against a feature.
In other words, for each visible feature:
- find whether one or a few clusters are significantly different from others. ("Only Cluster 5 has a significantly high Sikh %")
- find whether one or a few clusters have the highest value for a feature (e.g. "Cluster 5 has the highest electricity %")
Similarly, for each visible cluster, find whether one or a few features are significantly different from others. ("Cluster 3 is distinguished by a low rural % and high literacy %")
String these insights into a summary paragraph and display it.
Note: if the user hides a cluster or feature. remove those from the summary.