Reorder factors by increasing frequency

问题

How do I reorder factor-valued columns by frequency - in increasing order?

While the forcats package provides an explicit way to reorder a factor based on its frequency (fct_infreq()), it does so in decreasing frequency order. I need the reverse order of the factor frequency/counts.

E.g.

library(forcats)
set.seed(555)
df <- data.frame(x=factor(sample(as.character(1:10), 100, replace=TRUE)))
table(df$x)

1 10   2  3  4  5  6  7  8  9 
9 10  12 14 10 10  5 12  8 10 

levels(fct_infreq(df$x))

[1] "3"  "2"  "7"  "10" "4"  "5"  "9"  "1"  "8"  "6" 

Is there a simple way to flip the ordering so that the least frequent factor ("6") is first and the most frequent ("3") last?

回答1:

This can be done simply by using fct_rev as follows:

levels(fct_rev(fct_infreq(df$x)))

[1] "6"  "8"  "1"  "9"  "5"  "4"  "10" "7"  "2"  "3" 

回答2:

Or, you can do it in base R by sorting and resetting the levels.

xLev = names(table(df$x))[order(table(df$x))]
df$x = factor(df$x, levels=xLev)
table(df$x)
 6  8  1 10  4  5  9  2  7  3 
 5  8  9 10 10 10 10 12 12 14 

回答3:

with(data.frame(table(df$x)), setNames(sort(Freq), Var1[order(Freq)]))
# 6  8  1 10  4  5  9  2  7  3 
# 5  8  9 10 10 10 10 12 12 14 

来源：https://stackoverflow.com/questions/45846965/reorder-factors-by-increasing-frequency