Research Article

Multiple Naïve Bayes Classifiers Ensemble for Traffic Incident Detection

Table 4

Experimental Results of NB, Five Rules and NBTree Based as Applied to the AYE Dataset (The performances are presented in the form average ± variance).

AlgorithmDRFARMTTDCRAUCKappaMAERMSEEC

Naïve Bayes classifier0.7112 ± 0.003480.0499 ± 8.16E − 041.394 ± 0.088690.9094 ± 0.001110.8663 ± 4.39E − 040.6247 ± 9.14E − 040.16769 ± 3.07E − 040.57895 ± 5.31E − 040.90854 ± 1.83E − 04

Product rule ensemble0.6457 ± 0.001270.0557 ± 2.40E − 061.8685 ± 1.241390.8732 ± 1.64E − 060.8619 ± 4.85E − 070.7479 ± 3.58E − 050.16690 ± 1.04E − 060.57775 ± 3.11E − 060.90895 ± 3.68E − 07

Sum rule ensemble0.7923 ± 5.57E − 050.0500 ± 6.73E − 071.4384 ± 0.638680.8774 ± 3.55E − 070.8667 ± 3.75E − 070.7439 ± 1.29E − 040.16739 ± 4.94E − 070.57861 ± 1.47E − 060.90866 ± 1.75E − 07

Max rule ensemble0.7869 ± 1.79E − 040.0500 ± 2.21E − 061.4571 ± 0.722720.8771 ± 1.25E − 060.8663 ± 2.02E − 070.7511 ± 4.12E − 050.16803 ± 4.96E − 060.57969 ± 1.46E − 050.90828 ± 1.77E − 06

Min Rule ensemble0.7960 ± 1.25E − 040.0498 ± 2.76E − 061.4646 ± 0.735390.8781 ± 1.10E − 060.8667 ± 1.14E − 060.6052 ± 1.86E − 060.16638 ± 3.25E − 060.57684 ± 9.83E − 060.90926 ± 1.15E − 06

MV rule ensemble0.6246 ± 1.15E − 050.0572 ± 2.29E − 061.8358 ± 1.191240.8720 ± 3.84E − 070.7837 ± 1.18E − 060.7341 ± 5.31E − 050.16687 ± 2.27E − 060.57769 ± 6.87E − 060.90897 ± 8.03E − 07

NBTree classifier0.7275 ± 3.25E − 40.0250 ± 1.85E − 51.2103 ± 0.103830.9031 ± 3.52E − 50.8512 ± 1.04E − 40.7085 ± 3.26E − 40.1711 ± 6.42E − 50.49189 ± 2.62E − 40.90553 ± 2.06E − 5