After replacing zeros with NaNs, which feature has the highest number of missing values?

After replacing zeros with NaNs, which feature has the highest number of missing values?: a task in SmolDataEnvs-harbor-train-sandbox: 📊 SmolDataEnvs: Harbor (train) (Harbor dataset). Files (in /home/user/input, no subfolders): - diabetes.csv