Both conventional and machine learning-based density functional approximations (DFAs) have emerged as versatile approaches for virtual high-throughput screening and chemical discovery. To date, however, no single DFA is universally accurate for different chemical spaces. This DFA sensitivity is particularly high for open-shell transition-metal-containing systems, where strong static correlation may dominate. With electron density fitting and transfer learning, we build a DFA recommender that selects the DFA with the lowest expected error in a system-dependent manner. We demonstrate this recommender approach on the prediction of vertical spin-splitting energies (i.e., the electronic energy difference between the high-spin and low-spin state) of challenging transition metal complexes. This recommender yields relatively small errors (i.e., 2.1 kcal/mol) for transition metal chemistry and captures the distributions of the DFAs that are most likely to be accurate.