Table 2.

Accuracy and 95% confidence intervals for Random Forest prediction with the top 16 features selected using Recursive feature elimination (RFE).

	CSD17
	Continent	City
AMR	0.55 (0.48,0.61)	0.46 (0.4,0.52)
proGenomes	0.81 (0.76,0.86)	0.71 (0.65,0.76)

The top predicted cities using AMR count taxa are Stockholm, Taipei, and New York City (balanced accuracy 88, 88, and 85%, respectively) while the top predicted with proGenomes count data are Kuala Lumpur, London, and New York City with almost perfect accuracy.