Imagine a Walkable City: Physical activity and urban imageability across 19 major cities

EPJ Data Science

Table 3 Linear regressions that predict physical activity from traditional proxies for physicial activity (lnGDP, clemency, greenery, walkability, and landmarks), and from our street entropy metric. Z-scored transformed predictors are noted with ‘z’, and significant predictors with p values < 0.05 are marked in bold. The best predictive model (\(M_{IG+E+W+L}\)) suggests that both street entropy and the interaction between lnGDP and clemency are good predictors of physical activity

Predictor	β	std. error	p-value
\(\boldsymbol{M_{lG+C}}\): \(\boldsymbol{\mathit{Adj} R^{2} = 0.14}\), Durbin-Watson = 0.77, AIC = 54.51
Intercept	4.86	11.87	0.68
lnGDP	−0.41	0.85	0.63
Clemency	−0.04	0.12	0.7
lnGDPxClemency	−0.74	0.65	0.27
\(\boldsymbol{M_{lG+C+EU}}\): \(\boldsymbol{\mathit{Adj} R^{2} = 0.26}\), Durbin-Watson = 0.88, AIC = 52.29
Intercept	−2.47	11.68	0.83
lnGDP	0.11	0.83	0.89
Clemency	−0.04	0.11	0.72
lnGDPxClemency	−0.57	0.61	0.36
EU	1.09	0.58	0.08
\(\boldsymbol{M_{lG+C+E+G}}\): \(\boldsymbol{\mathit{Adj} R^{2} = 0.37}\), Durbin-Watson = 2.33, AIC = 50.24
Intercept	12.05	11.37	0.31
lnGDP	−0.93	0.81	0.27
Clemency	0.01	0.11	0.89
lnGDPxClemency	−1.33	0.62	0.05
z-street Entropy	0.64	0.22	0.01
z-Greenery	−1.20	3.11	0.7
lnGDPxz-Greenery	0.09	0.24	0.71
\(\boldsymbol{M_{lG+C+E}}\): \(\boldsymbol{\mathit{Adj} R^{2} = 0.46}\), Durbin-Watson = 1.32, AIC = 46.55
Intercept	10.17	9.61	0.31
lnGDP	−0.79	0.68	0.26
Clemency	0.004	0.09	0.96
lnGDP×Clemency	−1.26	0.54	0.03
z-street Entropy	0.61	0.19	0.00
\(\boldsymbol{M_{lG+E+W+L}}\): \(\boldsymbol{\mathit{Adj} R^{2} = 0.52}\), Durbin-Watson = 1.19, AIC = 44.69
Intercept	10.17	9.61	0.31
lnGDP	−0.58	0.36	0.13
lnGDP×Clemency	−0.97	0.37	0.02
z-street Entropy	0.57	0.18	0.00
Walkability	1.60	1.87	0.40
Landmarks	−5.72	5.36	0.30