The compute partials should, for now, be localized to inside the primary function. We don't use this function otherwise. We could hypothetically use one later to help with supplemental projections (of truly external data). Additionally, it might be worth dropping the barycentric scores, as this isn't what we typically visualize. Rather, this shows that the procedure is barycentric. But this depends on how we want to visualize...
To note: this might impact if/how we bootstrap, but that's for another issue.
The compute partials should, for now, be localized to inside the primary function. We don't use this function otherwise. We could hypothetically use one later to help with supplemental projections (of truly external data). Additionally, it might be worth dropping the barycentric scores, as this isn't what we typically visualize. Rather, this shows that the procedure is barycentric. But this depends on how we want to visualize...
To note: this might impact if/how we bootstrap, but that's for another issue.