Simple: Remove all attributes for which all rows are empty/null/missing/...
Better: List all attributes for which all rows are empty and allow selection of which ones to remove.
This would really help to reduce the number of columns when normalizing data sets from various sources. My continuous need is for normalizing addresses from various county, city and open sources.
When you have a couple hundred thousand rows, it is hard to figure out which attributes actually add value via the AttributeRemover or AttributeKeeper.
