Figure 5

Random Forest Classification Models using Three Ambient Modalities. Random Forest confusion matrices comparing the species level classification accuracy of electrical diathermy-based REIMS (handheld bipolar and monopolar platform) with ALDI-MS using the 50 to 500 m/z range (fatty acids and low molecular weight metabolites) and 600 to 1000 m/z range (phospholipids). Global accuracies for each as shown in matrix notations: (a) 96.1%; (b) 89.4%; (c) 98.8%; (d) 98.8%; (e) 98.8%; (f) 99.4%.