Here's a training subsample - note that Xbtm ranges through a couple of orders of magnitude (like depth). Should we be log transforming it, too?
class MLD Sbtm SSS
presence :1840 Min. : 1.654 Min. :28.35 Min. :27.70
background:3613 1st Qu.: 4.165 1st Qu.:32.40 1st Qu.:30.60
Median : 4.686 Median :34.00 Median :30.69
Mean : 4.737 Mean :33.54 Mean :30.83
3rd Qu.: 5.114 3rd Qu.:34.66 3rd Qu.:30.98
Max. :12.426 Max. :35.15 Max. :33.79
SST Tbtm U V
Min. :14.42 Min. : 2.109 Min. :-0.1860192 Min. :-0.1549941
1st Qu.:18.76 1st Qu.: 6.029 1st Qu.:-0.0039864 1st Qu.:-0.0048306
Median :19.95 Median : 7.419 Median :-0.0009787 Median :-0.0008589
Mean :20.25 Mean : 7.599 Mean :-0.0063102 Mean :-0.0050383
3rd Qu.:21.83 3rd Qu.: 8.372 3rd Qu.: 0.0010610 3rd Qu.: 0.0015496
Max. :24.97 Max. :24.211 Max. : 0.0325439 Max. : 0.0229367
Xbtm depth
Min. :2.874e-05 Min. : 5.00
1st Qu.:8.552e-04 1st Qu.: 62.72
Median :1.700e-03 Median : 123.30
Mean :5.726e-03 Mean : 379.28
3rd Qu.:3.573e-03 3rd Qu.: 202.39
Max. :1.490e-01 Max. :4081.49
