Following the footsteps of giants: modeling the mobility of historically notable individuals using Wikipedia

EPJ Data Science

Table 4 Models’ performances. Results obtained for the five models on predicting the number of destinations, the radius of gyration, and the distribution of the length of the migration jumps. The metrics used are the adjusted-\(R^{2}\), the Pearson correlation coefficient, ρ, between models and data, the Kullback–Leibler distance (K-L dist), and the first Wasserstein distance (Wass. dist)

Model	\(\text{adj-}R^{2}\)	Pearson ρ	K-L dist	Wass. dist
Radius of gyration
Pop-notable-multilevel	0.2414 ± 0.0027	0.962^∗∗∗	0.00554 ± 0.00004	0.000100 ± 2e − 7
Pop-multilevel	−0.2004 ± 0.0034	0.953^∗∗∗	0.00655 ± 0.00005	0.000125 ± 1.e − 7
Notable-multilevel	−0.6849 ± 0.0041	0.947^∗∗∗	0.00836 ± 0.00005	0.000139 ± 1e − 7
Notable-singlelevel	−1.0249 ± 0.0048	0.923^∗∗∗	0.01006 ± 0.00006	0.000143 ± 1e − 7
Random-singlelevel	−2.2673 ± 0.0054	0.886^∗∗∗	0.01559 ± 0.00009	0.000173 ± 1e − 7
Different destinations
Pop-notable-multilevel	0.9547 ± 0.0004	0.978^∗∗∗	0.0649 ± 0.002	0.0150 ± 0.0001
Pop-multilevel	0.9612 ± 0.0004	0.981^∗∗∗	0.0561 ± 0.001	0.0154 ± 0.0001
Notable-multilevel	0.9619 ± 0.0003	0.982^∗∗∗	0.0570 ± 0.001	0.0155 ± 0.0001
Notable-singlelevel	0.9624 ± 0.0003	0.982^∗∗∗	0.0623 ± 0.002	0.0159 ± 0.0001
Random-singlelevel	0.9606 ± 0.0004	0.982^∗∗∗	0.0724 ± 0.002	0.0163 ± 0.0001
Length of migration jumps
Pop-notable-multilevel	0.5104 ± 0.0019	0.982^∗∗∗	0.00533 ± 0.00005	0.000080 ± 1e − 7
Pop-multilevel	0.2249 ± 0.0023	0.974^∗∗∗	0.00686 ± 0.00005	0.000099 ± 1e − 7
Notable-multilevel	−0.0640 ± 0.0029	0.967^∗∗∗	0.00795 ± 0.00005	0.000109 ± 1e − 7
Notable-singlelevel	−0.2192 ± 0.0029	0.962^∗∗∗	0.00790 ± 0.00006	0.000112 ± 1e − 7
Random-singlelevel	−0.8313 ± 0.0034	0.947^∗∗∗	0.01265 ± 0.00006	0.000131 ± 1e − 7