Forum: help
Monitor Forum | 
Start New Thread
| RE: kernel density [ Reply ] By: Brigitta Toth on 2015-02-04 09:49 | [forum:41870] |
|
Dear Julien! Thanks a lot for your answer. OK, I will try to use kde2d() on clay and sand fractions to get closer to the the meaning of numbers shown on the isolines. Kind greetings, Brigitta |
|
| RE: kernel density [ Reply ] By: Julien Moeys on 2015-02-03 23:06 | [forum:41867] |
|
Hi Brigitta Sorry (again!) for the late reply. Not much time for my R packages at the moment. This kernel density estimate was made more for exploratory, visual, analysis than for complex calculations. I haven't used it for a long time. The function TT.kde2d() is based essentially on kde2d() (from the pacakge MASS). It simply transforms the CLAY-SILT-SAND coordinates into x-y coordinates, and then runs kde2d() on the transformed data. The result is a grid of density estimates (that can be plotted with image for instance). It is also similar to the function density() (but in 2D). You could also obtain similar results by applying kde2d() on two of the three texture classes. I must admit I don't know so well what is the "unit" of density estimates! I assume it must be relative units. The range of density obtained probably also depends on how spread (or how concentrated) your data are, I suppose. But interesting question! /Julien |
|
| kernel density [ Reply ] By: Brigitta Toth on 2014-07-18 16:13 | [forum:41866] |
|
Dear Julien, I would like to ask your help in using kernel density. This is very helpful, when we would like to plot hundreds of data, but for me it is not clear how values shown on contour plots are computed. For my data I got similar values to yours shown in soiltexture_vignette.pdf, it is usually between: 0.0001 and 0.0007. Could you please give the formula of it? With kind regards, Brigitta Toth |
|
