Fig. 4From: Fingerprinting cities: differentiating subway microbiome functionalityDistribution of top-match scores from final-SVM. The columns from the left are: known set, random set, SAND set, Ilorin samples from unknown set, Lisbon samples from unknown set, Boston samples from unknown set and mix set. The black dash line indicates 0.65, the cutoff below which the samples are likely to be random, i.e., the sample is not from any of the eight cities with which we trained our modelBack to article page