Page 49 - Read Online

Page 6 of 20 Zhu et al. J. Mater. Inf. 2025, 5, 8 https://dx.doi.org/10.20517/jmi.2024.76

Table 1. List of features used
Number Element properties Number Element properties
1 Electron binding energies 35 Gs Est Bcc latcnt
2 Nearest neighbor distance 36 Gs Est Fcc latcnt
3 Atomic concentration 37 Gs mag moment
4 Configuration energy 38 Gs volume Per
5 Cohesive energy 39 Hhi_p
6 Gram atomic volume 40 Hhi_r
7 Molar mass 41 Heat capacity mass
8 Screening percentage 42 Heat capacity molar
9 Enthalpy of vaporization 43 Icsd volume
10 Enthalpy of fusion 44 Heat of formation
11 First ionization energy 45 Lattice constant
12 Atomic radius 46 Mendeleev number
13 Atomic radius rahm 47 Melting point
14 Atomic volume 48 Molar volume
15 Atomic weight 49 Num unfilled
16 Boiling point 50 Num valance
17 Bulk modulus 51 Num d unfilled
18 C6 Gb 52 Num d valence
19 Covalent radius cordero 53 Num f unfilled
20 Covalent radius pyykko 54 Num f valence
21 Covalent radius pyykko double 55 Num p unfilled
22 Covalent radius pyykko triple 56 Num p valence
23 Covalent radius slater 57 Num s unfilled
24 Density 58 Num s valence
25 Dipole polarizability 59 Period
26 Electron negativity 60 Specific heat
27 Electron affinity 61 Thermal conductivity
28 En allen 62 Vdw radius
29 En ghosh 63 Vdw radius alvarez
30 En pauling 64 Vdw radius Mm3
31 First ion En 65 vdw_radius_uff
32 Fusion enthalpy 66 Sound velocity
33 Gs bandgap 67 Polarizability
34 Gs energy

and training samples. RF consists of multiple decision trees, each trained on different subsets of data and
features. The final prediction is obtained by averaging the outputs or majority voting of all the trees, offering
high robustness and noise resistance. XGBoost is an optimized implementation of gradient boosting
decision tree, known for its precision and efficiency. It minimizes prediction errors by combining multiple
weak learners with weighted contributions, and is capable of handling high-dimensional data and missing
values.

The main idea in this study is to represent the crystal structure as graph structure data with nodes
corresponding to atoms and edges corresponding to chemical bonds. The entire graph and the
corresponding adjacency matrix are constructed based on the locations and connectivity of atoms in the
crystal. The properties of the nodes in the graph are represented by physical and chemical properties
inherent to the atoms, such as atomic concentration, atomic radius, and boiling point. Mathematically, let