An Optimized Machine Learning and Big Data Approach to Crime Detection

Table 3

Performance evaluation: San Francisco Dataset.

Classifier	15 days average			7 days average			2 days average
	Accuracy	Precision	Recall	Accuracy	Precision	Recall	Accuracy	Precision	Recall
NB	95	94	95.91	92.5	93.68	90.8	90.5	89.98	90.81
NB (with proposed features)	97.5	97.97	97	94.5	91.91	96.8	92.5	91.91	92.85
RF	93.5	94.05	93.13	89.5	87.87	90.6	90	87.87	91.57
RF (with proposed features)	97	96.03	97.97	93	88.11	97.8	91.5	88.11	94.68
kNN	95	93	96.87	94	92.85	94.79	90	89.21	91
kNN (with proposed features)	95.5	95.95	95	94	92.92	94.84	91.5	92.92	90.19
SVM	96.5	97	96.03	91.5	91.57	90.62	90	91.57	87.87
SVM (with proposed features)	97	98	96.07	92.5	97.8	87.25	91.5	95.69	87.25