@KirkegaardEmil https://quid.com/feed/how-quid-uses-deep-learning-with-small-data … example of the sort of approach I mean fpr handling your surnames by regression on n-grams.
-
That *is* surprising. Do even surnames, marking family history/class/genetics, achieve CV performance like 29%?
Unfortunately, I don't have surnames. However, it's possible to estimate them by scraping all the first x last name numbers.