Presumably if you took the top 100, and manually merged then into the top 25? 50? job "categories", you'd have a more stable grouping and more interesting time-

Presumably if you took the top 100, and manually merged then into the top 25? 50? job "categories", you'd have a more stable grouping and more interesting time-series statistics.