Based on visual analysis, which feature pair provides better class separation for clustering purposes?

Based on visual analysis, which feature pair provides better class separation for clustering purposes?: a task in SmolDataEnvs-harbor-train-sandbox: 📊 SmolDataEnvs: Harbor (train) (Harbor dataset). Files (in /home/user/input, no subfolders): - Iris.csv - database.sqlite