dplyr examples: mutate + group

class: center, middle, inverse, title-slide

.title[
# dplyr examples: mutate + group_by
]
.author[
### Heike Hofmann
]

---

class: inverse, middle
# FBI data

---
class: inverse, middle
# `group_by` and `mutate`

---

# Working with `fbi`

`fbi` data included in the `classdata` package

upgrade the package: `remotes::install_github("heike/classdata")`

```r
library(tidyverse)
library(classdata)
data("fbi", package="classdata")
tail(fbi)
```

```
## # A tibble: 6 × 8
##   state       state_id state_abbr  year population type      count violent_crime
##   <chr>          <int> <chr>      <int>      <int> <chr>     <int> <lgl>        
## 1 Puerto Rico       43 PR          2020    3159343 robbery    1177 TRUE         
## 2 Puerto Rico       43 PR          2020    3159343 aggravat…  3342 TRUE         
## 3 Puerto Rico       43 PR          2020    3159343 burglary   2952 FALSE        
## 4 Puerto Rico       43 PR          2020    3159343 larceny    8311 FALSE        
## 5 Puerto Rico       43 PR          2020    3159343 motor_ve…  1978 FALSE        
## 6 Puerto Rico       43 PR          2020    3159343 arson        NA FALSE
```

---
class: inverse
# Your turn

For this your turn use the `fbi` data from the `classdata` package

- Use `mutate` to introduce a variable `rate` into the `fbi` data

- Use `mutate` to reorder types of crimes by (median) rate.

- Plot crime rates by type in side-by-side boxplots. Medians of the boxplots should be ordered.

---

#  `group_by` and `mutate`

Introduce a ranking by rate for each type of crime.

```r
fbi <- fbi %>% group_by(type) %>% mutate(
  rate = count/population*100000,
  rank = rank(rate) # ranks from lowest rate to highest rate
)
fbi %>% filter(rank == 1) %>% select(type, state, year, rate)
```

```
## # A tibble: 8 × 4
## # Groups:   type [8]
##   type                state         year    rate
##   <chr>               <chr>        <int>   <dbl>
## 1 robbery             North Dakota  1997   6.40 
## 2 aggravated_assault  North Dakota  1983  31.3  
## 3 homicide            North Dakota  1994   0.157
## 4 motor_vehicle_theft Vermont       2016  29.5  
## 5 rape_legacy         Puerto Rico   2013   0.723
## 6 rape_revised        Puerto Rico   2013   0.946
## 7 burglary            Puerto Rico   2020  93.4  
## 8 larceny             Puerto Rico   2020 263.
```

---
class: inverse
# Your turn

For this your turn use the `fbi` data from the `classdata` package 
- Introduce a ranking by rate for states for each type of crime and year.

- Focus on the top states. Find a visual that shows how often each state managed to take the top spot since 1961.

![](03_dplyr-examples_files/figure-html/unnamed-chunk-3-1.png)
---

# Keywords: which function is it?

- `mutate`: introduce, replace, reorder, ...

- `summarise`: calculate, average, summary, ...

- `group_by`: for each, across, ...

- `filter`: exclude, only consider, ...

---

# Avoiding potential traps

- using the $ notation in tidyverse can lead to strange behavior and error messages

- don't forget to save statements back into the dataset (`mutate`, `arrange`) or  new data objects (`summarise`, `filter`)

- when using the pipe `%>%`: what is output from lhs, first parameter on rhs?