Crime event prediction with dynamic features

EPJ Data Science

Table 5 Metrics comparison with and without dynamic features for Brisbane (“wod” and “wd” represent without and with dynamic features respectively)

Time interval	Theft		Drug Offence		Assault		Fraud		Unlawful Entry		Traffic Offence
Time interval	wod	wd	wod	wd	wod	wd	wod	wd	wod	wd	wod	wd
(a) Accuracy%
[9–12)	96.14	95.41	90.6	98.57	96.55	98.11	96.65	94.88	86.78	95.8	90.28	97.37
[12–15)	96.57	96.86	90.93	98.68	96.61	98.03	96.54	96.37	86.78	95.8	90.14	96.67
[15–18)	94.4	94.4	91.2	98.75	96.59	98.43	97.98	94.34	89.86	96.69	90.1	96.4
[18–21)	93.63	96.28	90.73	97.79	96.45	98.88	97.56	95.04	87.54	97.11	89.55	97.75
[21–24)	92.88	97.96	90.65	98.57	96.45	98.3	97.54	95.6	86.45	95.84	90.01	97.19
(b) F-1 Score%
[9–12)	98.03	97.65	95.07	98.78	97.24	98.0	98.29	97.37	92.92	97.86	94.89	98.67
[12–15)	98.25	98.4	95.25	98.84	97.28	98.0	98.24	98.15	92.92	97.86	94.81	98.3
[15–18)	97.1	97.1	95.33	98.88	97.27	98.2	98.88	97.09	94.67	98.32	94.8	98.17
[18–21)	96.76	98.11	95.14	98.88	97.19	98.78	98.76	97.46	93.36	98.53	94.48	98.87
[21–24)	96.3	98.97	95.09	98.28	97.19	98.15	98.76	97.75	92.73	97.88	94.74	98.56