Closed fmsabatini closed 3 years ago
@lenjon wrote in Issue: https://github.com/fmsabatini/sPlotOpen_Manuscript/issues/106
By the way, shall we also provide the pca3.RData object for free together with the code you wrote to generate that figure? That would be nice too :)
I also wonder if we could make an additional panel (or standalone figure) to display the spatial projection of PCA1 and PCA2, just that people have a sense of the meaning of these two PCA axes. What do you think?
We discussed the issue whether providing the environmental data, and we are inclined not to do so. @lgzrf Even if data from CHELSA and SOILGRIDS are shared under CC0, which would allow us to do so, these data are evolving over time, and providing a screenshot of these data would influence the users of sPlotOpen to stick to this outdated environmental data. This is relevant, since the whole resampling procedure was done 3 years ago, and both CHELSA and SOILGRIDS have been updated since then.
Since we're releasing the geographical coordinates, it will be easier for user to match the plots to whatever global environmental layer they prefer, anyways.
@lenjon, sure we could provide the PC1-PC2 values for each plot. But then we will also need to report enough data to allow the users to interpret these axes, I'm talking about the loadings and\or correlations b\w each variable and each PC, for instance using a biplot. Together with the graphs of the spatial distribution of PC1-PC2 values, wouldn't this end up taking too much space?
Data papers in GEB are limited to 2000 words (I have no idea where we stand, currently) and 2 Figures. Maybe the solution is to make a dedicated appendix only for PCA. Appendices are not allowed by Scientific data. I think they are in GEB, but I'm not 100% sure.
Check this out @lengyelat @lenjon For the appendix on PCA
Hi Francesco,
This figure with the two maps of the PC1 and PC2 axes is perfect and can indeed go in the appendices together with a biplot to show the laodings and correlations of the 30 variables along both PC1 and PC2, also mentioning briefly the amount of variance captrured by each of the two PC axes. This together with the two maps should be fair enough for the curious reader who would like to use our PCA outputs. Thus, in addition to providing the R code to run the PCA and prepare the raster layers of the two PCA axes (+ code of your very nice figures to display it), we could also provide the data for PC1 and PC2, basically the two global raster layers. I think it is a very nice and complementary addition to sPlotOpen and actually part of it since this is background environmental space to build sPlotOpen. Thus, it would make sense to provide this data layers (PC1 + PC2 grids) too to the user. Besides, this is also our own product and if we do not provide teh raw data to generate PC1 and PC2 (these data are free and available) we are at least providing out own product to then represent sPlotOpen within its sampling design space :) I could see people using sPlotOpen and wanting to display the data within the PC1-PC2 space :) I think this is super useful.
Best,
Jonathan
Jonathan Lenoir Chargé de Recherche CNRS
Université de Picardie Jules Verne Ecologie et Dynamiques des Systèmes Anthropisés UMR 7058 CNRS-UPJV 1 Rue des Louvels 80000 AMIENS FRANCE
http://www.u-picardie.fr/edysan/_listing-personnel/jonathan-lenoir/ http://scholar.google.com/citations?user=Xx5 http://scholar.google.com/citations?user=Xx52nH4AAAAJ2nH4AAAAJ http://scholar.google.com/citations?user=Xx52nH4AAAAJ http://www.researchgate.net/profile/Jonathan_Lenoir/ http://jonathanlenoir.wordpress.com/
Le mar. 8 déc. 2020 à 19:51, fmsabatini notifications@github.com a écrit :
Check this out @lengyelat https://github.com/lengyelat @lenjon https://github.com/lenjon For the appendix on PCA [image: figure4] https://user-images.githubusercontent.com/51127026/101527626-8980f000-398e-11eb-8746-fae2639e3951.png
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/fmsabatini/sPlotOpen_Manuscript/issues/120#issuecomment-740842851, or unsubscribe https://github.com/notifications/unsubscribe-auth/AELUDUUDSC3BOK7IEODKSVDSTZYSFANCNFSM4UCDX6HA .
I agree, this is a very informative pair of maps.
The values of the PC1-PC2 are not part of the header file.
114