[R] S-mode PCAs

Daniel Vecellio djvecellio at tamu.edu
Tue Jun 13 22:26:04 CEST 2017


Hi all,

I have a file of average SWE observations for 40 years at over 4,000 points
and am attempting to do a spatiotemporal analysis of the data using PCA,
much like this paper did using snow depth:
http://journals.ametsoc.org/doi/pdf/10.1175/1520-0442%281998%29011%3C0856%3ATCIRWS%3E2.0.CO%3B2

I have followed the code in the link below by my loadings are far too small
(For example, the paper listed above had loading above 0.5 and below -0.5
while my loadings for PC1 go between 0.02 and -0.03).

https://stackoverflow.com/questions/41022927/principal-component-analysis-pca-of-time-series-data-spatial-and-temporal-pat

I have standardized the data in the prcomp function so that a correlation
matrix is used in my study, just like the paper above. Does anyone have any
idea why my loadings might be so off and any suggestions regarding the PCA
functions in R to help solve this problem?

Thanks in advance,
Dan

-- 

*Daniel J. Vecellio*

*Ph.D. Student - Climate Science Lab, **Department of Geography, **Texas
A&M University*

	[[alternative HTML version deleted]]



More information about the R-help mailing list