Factor

---

Character to Factor

How to transform character vector into factor?

Here is an example vector of character class -

> gender_char <- c("female","female","female","male",
+                  "female","male","male","female","other")
> class(gender_char)

[1] "character"

Transform to factor using factor() -

> gender_fact <- factor(gender_char)
> class(gender_fact)

[1] "factor"

> levels(gender_fact)

[1] "female" "male"   "other" 

droplevels()

How to drop unused levels from a factor?

Let’s see how levels change after taking a subset of the data set -

> gender_subset <- gender_fact[1:4]
> gender_subset

[1] female female female male  
Levels: female male other

Although other is not in the data set it is showing up in the levels. So droplevels() function needs to be used -

> gender_subset <- droplevels(gender_subset)
> gender_subset

[1] female female female male  
Levels: female male

Number to factor

How to transform numeric vector into factor?

> gender_num <- c(0,0,0,1,0,1,1,0)
> class(gender_num)

[1] "numeric"

> gender_num_fact <- factor(gender_num)
> class(gender_num_fact)

[1] "factor"

gl()

To create a sequence of factor type data `gl() function can be used -

> fac_seq <- gl(n = 2, k = 3)
> fac_seq

[1] 1 1 1 2 2 2
Levels: 1 2

> class(fac_seq)

[1] "factor"

We can specify the levels using the argument labels-

> fac_seq <- gl(n = 2, k = 3, labels = c("Control", "Treatment"))
> fac_seq

[1] Control   Control   Control   Treatment Treatment Treatment
Levels: Control Treatment

If we want the levels to be ordered then ordered argument can be used -

> gl(n = 2, k = 3, labels = c("January", "May"), ordered = T)

[1] January January January May     May     May    
Levels: January < May

Changing factor levels

How to change factor levels?

> levels(gender_num_fact) <- c('Female','Male')
> gender_num_fact

[1] Female Female Female Male   Female Male   Male   Female
Levels: Female Male

Ordered Levels

How to make ordered levels?

Levels are usually ordered ascending order of alphabets -

> mymonth <- c('Jan','Dec','Mar','May','Sep')
> mymonth

[1] "Jan" "Dec" "Mar" "May" "Sep"

> mymonth_fact <- factor(mymonth)
> mymonth_fact

[1] Jan Dec Mar May Sep
Levels: Dec Jan Mar May Sep

But we know this is not right since the months are not in correct order. ordered = T preserves the order of levels passed in levels argument -

> mymonth_ord_fact <- factor(mymonth, levels = month.abb, ordered = T)
> #"month.abb" - base R's constant vector. Another example = "month.name"
> mymonth_ord_fact

[1] Jan Dec Mar May Sep
12 Levels: Jan < Feb < Mar < Apr < May < Jun < Jul < Aug < Sep < ... < Dec

Let’s drop the unused levels -

> mymonth_ord_fact <- droplevels(mymonth_ord_fact)
> mymonth_ord_fact

[1] Jan Dec Mar May Sep
Levels: Jan < Mar < May < Sep < Dec

Factor to Numeric

We can see some values are encoded as factors in the following vector -

> raise <- factor(c(1200,1500,1500,1200,1500,1800,1500,1200))
> raise

[1] 1200 1500 1500 1200 1500 1800 1500 1200
Levels: 1200 1500 1800

But when we try to convert it into numeric type, we’ll get the 1,2,3… .

> as.numeric(raise)

[1] 1 2 2 1 2 3 2 1

To get the numbers in output we need to convert it into strings first, and then into numeric -

> as.numeric(as.character(raise))

[1] 1200 1500 1500 1200 1500 1800 1500 1200