R Objects - Factors

class: center, middle, inverse, title-slide

# R Objects - Factors
## How to work with numbers and text together
### Melinda Higgins
### Emory University </br> 02/11/2021

---

# Numeric Data in R

.pull-left[

<img src="img/continuous_discrete.png" width="100%" />
]

.pull-right[
### CONTINUOUS - numeric (double)

```r
x <- c(1.12, 1.13, 4.23)
class(x)
```

```
## [1] "numeric"
```

### DISCRETE - integer

```r
y <- as.integer(c(1,2,3,4))
class(y)
```

```
## [1] "integer"
```
]

---

# Non-numeric Data in R

---

# Nominal Data

.left-column[
<img src="img/nominal.jpg" width="100%" />
]

### From character (text strings)

```r
animal <- 
  c("turtle","snail","butterfly",
    "turtle","butterfly")

class(animal)
```

```
## [1] "character"
```

```r
table(animal) %>% knitr::kable()
```

|animal    | Freq|
|:---------|----:|
|butterfly |    2|
|snail     |    1|
|turtle    |    2|

Notice that these are in alphabetical order

---

# Nominal Data as "Factor"

.pull-left[
Suppose on a survey each nurse selects the hospital where they work:

1. Emory Hospital (on Clifton Road)
2. Emory Midtown
3. Emory Decatur
4. Saint Joseph's

And we get the 7 responses of: 1, 1, 2, 3, 2, 2, 4

```r
hosp_choice <- c(1,1,2,3,2,2,4)
hospital <- 
  factor(hosp_choice,
         levels = c(1,2,3,4),
         labels = c("EHospital",
                    "EMidtown",
                    "EDecatur",
                    "St. Joe"))
```
]

.pull-right[
Notice that the

* `levels` corresponds to the response (choice) number and
* `labels` corresponds to the text of the response option.

```r
table(hospital) %>% knitr::kable()
```

|hospital  | Freq|
|:---------|----:|
|EHospital |    2|
|EMidtown  |    3|
|EDecatur  |    1|
|St. Joe   |    1|

Notice the responses are **in order by the response number** NOT alphabetical order.]

---

# Ordinal Data as "Factor"

.left-column[
<img src="img/ordinal.jpg" width="=100%" />
]

.right-column[
### Alphabetical Order is a Problem

Suppose we have "awful", "ok" and "awesome" as options for "how are you feeling today?" question.

```r
feeling <- 
  c("ok","ok","awesome","awful")
```

In a table, the default is alphabetical order:

```r
table(feeling) %>% knitr:: kable()
```

|feeling | Freq|
|:-------|----:|
|awesome |    1|
|awful   |    1|
|ok      |    2|
]

---

# Create Ordered Factor

List the text responses (levels) in the order you want and specify `ordered = TRUE`.

```r
feeling_ordf <- 
  factor(feeling,
         levels = c("awesome",
                    "ok",
                    "awful"),
*        ordered = TRUE)
```

Now the table is in a logical order

```r
table(feeling_ordf) %>% knitr:: kable()
```

|feeling_ordf | Freq|
|:------------|----:|
|awesome      |    1|
|ok           |    2|
|awful        |    1|

---

# Binary Data

.left-column[
<img src="img/binary.jpg" width="=100%" />
]

.right-column[
A variable is "binary" if it only has 2 values.

```r
animals <- c("shark","shark","dino","dino","shark")
table(animals) %>% knitr::kable()
```

|animals | Freq|
|:-------|----:|
|dino    |    2|
|shark   |    3|

Logical variables are also "binary".

```r
extinct <- animals == "dino"
table(extinct) %>% knitr::kable()
```

|extinct | Freq|
|:-------|----:|
|FALSE   |    3|
|TRUE    |    2|

]

---

# Binary Data - Use TRUE=1, FALSE=0

A neat feature of **TRUE/FALSE** logical variables is that because they convert to `0=FALSE, 1=TRUE` you can use:

* the `sum()` function to get the number (count) of **TRUE** and 
* use the `mean()` function to get the proportion of **TRUE**.

```r
sum(extinct)
```

```
## [1] 2
```

```r
mean(extinct)
```

```
## [1] 0.4
```

```r
# multiply by 100 to get percent TRUE
mean(extinct)*100
```

```
## [1] 40
```

Write a sentence in Rmarkdown with inline code using `` `r mean(extinct)*100` `` to write:

"The percentage of extinct animals was 40 %.