# -Dealing with Factors

 Factors are important in statistical modeling and are treated specially by modelling functions like `lm()` and `glm()`.

## #Creating, Converting & Inspecting Factors

```r
# create a factor string
gender <- factor(c("male", "female", "female", "male", "female"))
gender
## [1] male   female female male   female
## Levels: female male

# inspect to see if it is a factor class
class(gender)
## [1] "factor"

# show that factors are just built on top of integers
typeof(gender)
## [1] "integer"

# See the underlying representation of factor
unclass(gender)
## [1] 2 1 1 2 1
## attr(,"levels")
## [1] "female" "male"

# what are the factor levels?
levels(gender)
## [1] "female" "male"

# show summary of counts
summary(gender)
## female   male 
##      3      2
```

If we have a vector of character strings or integers we can easily convert to factors:

```r
group <- c("Group1", "Group2", "Group2", "Group1", "Group1")
str(group)
##  chr [1:5] "Group1" "Group2" "Group2" "Group1" "Group1"

# convert from characters to factors
as.factor(group)
## [1] Group1 Group2 Group2 Group1 Group1
## Levels: Group1 Group2
```

## #Ordering, Revaluing, & Dropping Factor Levels

#### Ordering Levels: <a href="#ordering-levels" id="ordering-levels"></a>

```r
# when not specified the default puts order as alphabetical
gender <- factor(c("male", "female", "female", "male", "female"))
gender
## [1] male   female female male   female
## Levels: female male

# specifying order
gender <- factor(c("male", "female", "female", "male", "female"), 
                 levels = c("male", "female"))
gender
## [1] male   female female male   female
## Levels: male female
```

&#x20;Create ordinal factors with  `ordered = TRUE` argument

```r
ses <- c("low", "middle", "low", "low", "low", "low", "middle", "low", 
"middle",
    "middle", "middle", "middle", "middle", "high", "high", "low", 
    "middle",
    "middle", "low", "high")

# create ordinal levels
ses <- factor(ses, levels = c("low", "middle", "high"), ordered = TRUE)
ses
##  [1] low    middle low    low    low    low    middle low    
# middle middle
## [11] middle middle middle high   high   low    middle middle low high  
## Levels: low < middle < high

# you can also reverse the order of levels if desired
factor(ses, levels=rev(levels(ses)))
##  [1] low    middle low    low    low    low    middle low   
# middle middle
## [11] middle middle middle high   high   low    middle middle low    
# high  
## Levels: high < middle < low
```

#### Revalue Levels: <a href="#revalue-levels" id="revalue-levels"></a>

To recode factor levels I usually use the `revalue()` function from the `plyr`package.

```r
plyr::revalue(ses, c("low" = "small", "middle" = "medium", "high" = 
"large"))
##  [1] small  medium small  small  small  small  medium small  
medium medium
## [11] medium medium medium large  large  small  medium medium 
small  large 
## Levels: small < medium < large
```

{% hint style="info" %}
&#x20;☛ *Using the `::` notation allows you to access the `revalue()` function without having to fully load the `plyr` package.*
{% endhint %}

#### Dropping Levels: <a href="#dropping-levels" id="dropping-levels"></a>

When you want to drop unused factor levels, use `droplevels()`:

```r
ses2 <- ses[ses != "middle"]

# lets say you have no observations in one level
summary(ses2)
##    low middle   high 
##      8      0      3

# you can drop that level if desired
droplevels(ses2)
##  [1] low  low  low  low  low  low  high high low  low  high
## Levels: low < high
```