Skip to main content
. 2021 Jun 8;9:e11573. doi: 10.7717/peerj.11573

Figure 8. Variable importance of spectral bands for clustering.

Figure 8

Bars represent relative importance of the spectral predictors for optimally classifying all clusters as calculated by a supervised random forest algorithm (Breiman, 2001). Specific variable importance for detecting cluster 1 is shown with open circles.