“…Intuitively and inevitably, successful and effective implementation of ML as predictive tools in electrocatalyst discovery entails a careful selection of input features (e.g., geometric, electronic, and energy properties) based on the established databases from experiments and in silico data from DFT calculations. Recently, the search for alternative electrocatalysts to Pt has predominantly focused on 2D materials such as graphdiyne (GDY), metal–nitrogen-carbon (M–N-C), MXene, and MoS 2 . − Sun et al have utilized ML as a complementary tool to DFT calculations in investigating GDY-based catalytic HER process. Specifically, the bagged-trees approach was adopted with six input features (i.e., mass number, active sites, d/f electrons, electronegativity, electron affinity, and ionization potential) to predict the HER performance, which yielded nearly identical results to DFT calculations (Figure c).…”