it's an iterative process
if you think they should have matched with some rows than yes more training is needed
z_minscore and z_maxscore columns would be null for rows which didn't have a match
attached is the csv file I used and output i got
which release of zingg are you using?
and it worked fine for me
I tried options = ClientOptions([ClientOptions.PHASE,"recommend",ClientOptions.COLUMN,"fname"])
Please try by giving ClientOptions.COLUMN instead of "--column"
