By comparing Table 4 and Table 6, and by comparing Table 5 and Table 7, the correlation coefficient of each group significantly improved (higher than the multiple liner regression model) in the GWR model. The predicted regression value of longevity by GWR with humidity, altitude, and per capita sea fish consumption, for each city, are illustrated in Figure 2.