Research Article

When Collective Knowledge Meets Crowd Knowledge in a Smart City: A Prediction Method Combining Open Data Keyword Analysis and Case-Based Reasoning

Table 11

Performance with depression dataset.

ClassifierClassOverall accuracyTP rateFP ratePrecisionRecall

SMONormal93.41320.9750.9330.9570.975
Serious0.0670.0250.1110.067
Overall0.9340.8930.9190.934
Overall (norm)0.5210.4790.5340.521

BayesNetNormal92.81440.9620.8000.9620.962
Serious0.2000.0380.2000.200
Overall0.9280.7660.9280.928
Overall (norm)0.5810.4190.5810.581

IBkNormal94.31140.9871.0000.9550.987
Serious0.0000.0130.0000.000
Overall0.9430.9560.9120.943
Overall (norm)0.4940.5070.4780.494

LogisticNormal91.91620.9470.6670.9680.947
Serious0.3330.0530.2270.333
Overall0.9190.6390.9350.919
Overall (norm)0.6400.3600.5980.640

C4.5Normal93.11380.9690.8670.9600.969
Serious0.1330.0310.1670.133
Overall0.9310.8290.9240.931
Overall (norm)0.5510.4490.5640.551

RipperNormal93.71260.9660.6670.9690.966
Serious0.3330.0340.3130.333
Overall0.9370.6380.9390.937
Overall (norm)0.6500.3500.6410.650

NRBNFNormal94.01200.9810.9330.9570.981
Serious0.0670.0190.1430.067
Overall0.9400.8920.9210.940
Overall (norm)0.5240.4760.5500.524

Crowd reasoningNormal94.61080.9720.6000.9720.972
Serious0.4000.0280.6000.400
Overall0.9360.5640.9490.936
Overall (norm)0.6860.3140.7860.686

Hybrid reasoningNormal92.51500.9400.4000.9800.940
Serious0.6000.0600.4000.600
Overall0.9190.3780.9440.919
Overall (norm)0.7700.2300.6900.770