i think the plotting interface actually makes a lot more sense to take a trained model as an argument. i'm fine removing the old interface (i can do in a subsequent PR)
Coverage increased (+0.1%) to 95.17% when pulling f5d114f1532923d8da2a52ee3121f3b28aeb3a55 on stphnma:plot_specific_groups into abb6d2a807318e5c398d1a6347a9f44bd1919811 on better:master.
👍 if you can fix the hound warnings
i think the plotting interface actually makes a lot more sense to take a trained model as an argument. i'm fine removing the old interface (i can do in a subsequent PR)