I have a multi domain data source, and I'm trying to visualize it. It contains many values, strings, timestamps, lat/long, numbers, labels etc.. Of course our display can only show a few dimensions (usually 2) at a time, so I have to programmatically select which attributes to show.
I'm trying to do this based upon attribute quality, let's say the lat/long attribute of the set is good if the elements are evenly spread on the map, not too clustered.. For a category label, the attribute is good if there aren't too many categories, or at least most elements fit in the first few categories..
If anyone grasped the point, I need to devise another few algorithms to compute data quality.. for example for float numbers, or timestamps..
Any idea is appreciated