在 R 中重新缩放，固定为百分位数

Question

I understand how to use rescale, but I'm unsure how I could modify it so rather than scaling the min and max (to a 0-100) scale, it scales the min to 0 and x percentile to 100.

Example:

Raw       0-100 scaling (wrong)   0-100 scaling (right [using 75th perecntile as max])
0         0                       0
1         10                      50
2         20                      100
10        100                     500

Current code is:

data$Predict_Party <- rescale(data$Predict_Party_Unscaled, to = c(0, 100))

Additional: How would you recommend I then determine the formula of the transformation that was used? Ie scaled = unscaled*x + y

Answer 1

Per helpful comment on the choice of algorithm I believe this is the answer.

library(scales)
raw <- c(0, 1, 2, NA, 10)
rescale(raw, to = c(0,100), from = c(0, quantile(na.omit(raw), .75, type = 1)))
#> [1]   0  50 100  NA 500
df <- data.frame(raw)
df$rescaled <- rescale(df$raw, to = c(0,100), from = c(0, quantile(na.omit(df$raw), .75, type = 1)))
df
#>   raw rescaled
#> 1   0        0
#> 2   1       50
#> 3   2      100
#> 4  NA       NA
#> 5  10      500