Table 2. Support vector machine (SVM) classification results (i.e., leave-one-out cross-validations of 112 permutations) of 16 common explosives, related compounds, and the control. The accuracy shown for each analyte represents the percentage of correctly identified analytes among 7 independent trials. Similarly, the misidentifications indicate which incorrect classifications were supplied by the algorithm. Consequently, the KClO₃ analytes should be considered a single group of analytes, and likewise, DMDNB and PETN.