view article
Figure 3
Scheme of the data analysis procedure. (a) Collection of azimuthally integrated SAXS/WAXS intensity curves Ii(q), for i = 1,...,r, where r is the number of I(q) points. (b) M2 is formed by selecting only the d intensities Ii(q) where inflection points occur. (c) Dimensionality reduction: principal component analysis is applied to M2 and the number of main variables is reduced to m principal components PCj, for j = 1,...,m. The optimal number of clusters n is evaluated and the signals are classified into n clusters. We assume n = 4 in this example. The scanned map is segmented according to the clustering results. (d) Estimation of a representative signal Sl(q) by selecting points that are located furthest from all centroids, for l = 1,...,n. Further details about each step can be found in the text.

Journal logoJOURNAL OF
APPLIED
CRYSTALLOGRAPHY
ISSN: 1600-5767
Volume 51| Part 5| October 2018| Pages 1378-1386
https://doi.org/10.1107/S1600576718011032
Follow J. Appl. Cryst.
Sign up for e-alerts
E-alerts
Follow J. Appl. Cryst. on Twitter
Twitter
Follow us on facebook
Facebook
Sign up for RSS feeds
RSS