Mean Absolute Error (MAE) for Property Prediction

id: mean_absolute_error_materials· bridge construct

Primary regression metric in Matbench: average absolute difference between predicted and DFT-computed material properties (eV/atom for energies, eV for band gaps, GPa for moduli). Lower is better; state-of-the-art models achieve ~0.02–0.05 eV/atom for formation energy.

cross-pax synthesis This construct appears in 2 pax. Agents can run analyses that span all of them — findings from each pax are directly comparable because they share this construct definition.

// usage

Used in 2 pax

F field Bartel Comp Materials

↑ positive relationship 4 findings

F field Machine learning for computational materials discovery

↑ positive relationship 4 findings

← back to construct index