我想知道是否可以在矩阵或数据帧上使用非标准化函数。我可以毫无疑问地在列上使用标准应用，但是不确定如何反转逻辑。这也已通过缩放功能完成。有什么建议吗？请参阅下面的reprex。
library(tibble)
library(cbar)
df <- tibble(
x = rnorm(10, 5, .1),
y = rnorm(10, 50, 1),
z = rnorm(10, 500, 20)
)
df
#> # A tibble: 10 x 3
#> x y z
#> <dbl> <dbl> <dbl>
#> 1 4.99 52.0 530.
#> 2 5.06 48.5 527.
#> 3 5.05 49.5 504.
#> 4 5.08 48.2 512.
#> 5 4.99 49.8 496.
#> 6 4.97 49.2 517.
#> 7 4.97 50.4 483.
#> 8 4.91 49.1 498.
#> 9 4.87 51.6 478.
#> 10 4.86 49.0 531.
scale(df)
#> x y z
#> [1,] 0.16214478 1.79905676 1.1749160
#> [2,] 1.10506079 -1.00798624 1.0366936
#> [3,] 0.98384421 -0.21586991 -0.1861718
#> [4,] 1.35024672 -1.18719295 0.2255299
#> [5,] 0.21101616 0.07949959 -0.6128166
#> [6,] -0.01764382 -0.40700291 0.4988026
#> [7,] -0.02722854 0.57290776 -1.2813415
#> [8,] -0.90900232 -0.51084700 -0.5015272
#> [9,] -1.38320530 1.47438778 -1.5696862
#> [10,] -1.47523268 -0.59695288 1.2156013
#> attr(,"scaled:center")
#> x y z
#> 4.973533 49.725633 507.582204
#> attr(,"scaled:scale")
#> x y z
#> 0.07522165 1.24812951 19.06825608
apply(df, 2, standardized)
#> x y z
#> [1,] 0.16214478 1.79905676 1.1749160
#> [2,] 1.10506079 -1.00798624 1.0366936
#> [3,] 0.98384421 -0.21586991 -0.1861718
#> [4,] 1.35024672 -1.18719295 0.2255299
#> [5,] 0.21101616 0.07949959 -0.6128166
#> [6,] -0.01764382 -0.40700291 0.4988026
#> [7,] -0.02722854 0.57290776 -1.2813415
#> [8,] -0.90900232 -0.51084700 -0.5015272
#> [9,] -1.38320530 1.47438778 -1.5696862
#> [10,] -1.47523268 -0.59695288 1.2156013
apply(scale(df), 2, destandardized, y_mu = colMeans(df), y_sd = apply(df, 2, sd))
#> Warning in y_hat * y_sd: longer object length is not a multiple of shorter
#> object length
#> Warning in y_hat * y_sd + y_mu: longer object length is not a multiple of
#> shorter object length
#> Warning in y_hat * y_sd: longer object length is not a multiple of shorter
#> object length
#> Warning in y_hat * y_sd + y_mu: longer object length is not a multiple of
#> shorter object length
#> Warning in y_hat * y_sd: longer object length is not a multiple of shorter
#> object length
#> Warning in y_hat * y_sd + y_mu: longer object length is not a multiple of
#> shorter object length
#> x y z
#> [1,] 4.985730 5.108861 5.061912
#> [2,] 51.104892 48.467535 51.019561
#> [3,] 526.342398 503.465942 504.032232
#> [4,] 5.075101 4.884231 4.990498
#> [5,] 49.989008 49.824858 48.960758
#> [6,] 507.245768 499.821369 517.093500
#> [7,] 4.971485 5.016628 4.877149
#> [8,] 48.591080 49.088029 49.099662
#> [9,] 481.206892 535.696208 477.651026
#> [10,] 4.862564 4.928630 5.064973
Created on 2018-11-08 by the reprex package (v0.2.0).