In STAT545-UBC-students/hw07-ziqiangt: Make Factors Less Annoying

Factors are a very useful type of variable in R, but they can also drive you nuts. Especially the "stealth factor" that you think of as character.

Can we soften some of their sharp edges?

Binding two factors via fbind():

library(foofactors)
a <- factor(c("character", "hits", "your", "eyeballs"))
b <- factor(c("but", "integer", "where it", "counts"))

Simply catenating two factors leads to a result that most don't expect.

c(a, b)

The fbind() function glues two factors together and returns factor.

fbind(a, b)

Often we want a table of frequencies for the levels of a factor. The base table() function returns an object of class table, which can be inconvenient for downstream work. Processing with as.data.frame() can be helpful but it's a bit clunky.

set.seed(1234)
x <- factor(sample(letters[1:5], size = 100, replace = TRUE))
table(x)
as.data.frame(table(x))

The freq_out() function returns a frequency table as a well-named tbl_df:

freq_out(x)

detect factors that should be character because # unique values = length

f_detect(factor(c("a", "b", "c","a")))
f_detect(factor(c("a", "b", "c","d")))

write a version of reorder() that uses desc() a la (d)plyr

f_reorder(factor(c("B", "A", "D")))

write a version of factor() that sets levels to the order in which they appear in the data, i.e. set the levels “as is”

f_set(factor(c("B", "A", "D")))

functions to write and read data frames to plain text delimited files while retaining factor levels; maybe by writing/reading a companion file?

# De
set.seed(1234)
df <- data.frame(
    kids = factor(c(1,0,1,0,0,0), levels = c(0, 1),
    labels = c("boy", "girl"))
  )

levels(df$kids)

x_write(df, "./df_x.csv", "./df_x.txt")
read_return <- x_read("./df_x.csv", "./df_x.txt")
levels(read_return$kids)

STAT545-UBC-students/hw07-ziqiangt documentation built on May 24, 2019, 7:53 a.m.