Skip to main content

Table 2 Model validation results of two experimental settings

From: Quantifying the economic impact of disasters on businesses using human mobility data: a Bayesian causal inference approach

   Evaluation metric Use of covariates
   No covariates \(x_{category}\) \(x_{specific}\)
Setting 1 Train MAPE 12.35 (±16.67) 10.50 (±14.03) 10.66 (±14.46)
Pearson R 0.539 (±0.222) 0.696 (±0.169) 0.626 (±0.136)
Test MAPE 8.568 (±14.37) 8.518 (±15.85) 8.888 (±15.30)
Pearson R 0.351 (±0.238) 0.354 (±0.239) 0.295 (±0.257)
Selected (%) 34.9 40.4 24.7
Setting 2 Train MAPE 0.229 (±0.257) 0.249 (±0.251) 0.257 (±0.252)
Pearson R 0.855 (±0.144) 0.742 (±0.145) 0.744 (±0.115)
Test MAPE 0.704 (±0.811) 0.475 (±0.612) 0.477 (±0.538)
Pearson R 0.420 (±0.189) 0.512 (±0.181) 0.466 (±0.183)
Selected (%) 40.3 25.1 34.6