Figure 2 Unsupervised learning methods, PCA, and k-means on the FeSeTe CITS data. (a) 150 × 150 pixel CITS data from the Z-channel before
the feedback is disabled for the IV spectroscopy. The inset shows the average IV for the data set. (b) Second PCA loading of the CITS data shown
in (a); the inset shows the eigenvector. (c) Third PCA loading of the CITS data shown in (a); the inset shows the eigenvector. (d) k-means clustering
results for a five-cluster case; the inset shows the mean IV for each cluster.