Social Media Meets Big Urban Data: A Case Study of Urban Waterlogging Analysis

Table 2

The results of baseline and transfer learning for five target cities.

Cities	Social view	Physical view	Total
GBRT
Beijing	0.745	0.613	0.815
Shanghai	0.732	0.589	0.712
Hangzhou	0.555	0.511	0.711
Mumbai	0.781	0.631	0.811
New Delhi	0.781	0.711	0.722
ANN (not multiview)
Beijing	0.742	0.675	0.802
Shanghai	0.715	0.688	0.803
Hangzhou	0.713	0.566	0.814
Mumbai	0.832	0.662	0.802
New Delhi	0.754	0.744	0.813
TrAdaBoost (instance-transfer)
Beijing	0.762	0.676	0.812
Shanghai	0.755	0.617	0.883
Hangzhou	0.733	0.555	0.824
Mumbai	0.832	0.722	0.852
New Delhi	0.724	0.784	0.862
Multiview transfer learning with autocoder (feature-transfer)
Beijing	0.753	0.638	0.852
Shanghai	0.755	0.687	0.863
Hangzhou	0.763	0.595	0.894
Mumbai	0.832	0.732	0.892
New Delhi	0.754	0.794	0.912