---
title: "Counts"
output: rmarkdown::html_vignette
fig.width: 8
fig.height: 5
vignette: >
  %\VignetteIndexEntry{Counts}
  %\VignetteEncoding{UTF-8}
  %\VignetteEngine{knitr::rmarkdown}
editor_options: 
  markdown: 
    wrap: 80
---

## Counts

A first step in many transportation safety analyses involves counting the number
of relevant crashes, fatalities, or people involved. `counts()` lets users
specify *what* to count, *where* to count them (rural/urban and/or in specified
states or regions), *who* to include, the *interval* over which to count
(annually or monthly), and factors *involved* in the crashes. It returns a
simple tibble that can be easily piped into `ggplot()` to quickly visualize
counts.

First we load the required libraries:

```{r, message=F}
library(dplyr)
library(ggplot2)
library(tidyr)
library(rfars)
```

### annual_counts

`rfars` includes `annual_counts`, a table of annual crash counts:

```{r}
rfars::annual_counts %>%
  filter(what == "crashes", involved == "any") %>%
  ggplot(aes(x=year, y=n)) +
  geom_col() +
  facet_wrap(.~source, nrow=1, scales = "free_y") +
  labs(title = "Total annual crashes by type (FARS = fatal, CRSS = general)", x=NULL, y=NULL) +
  theme_minimal()

rfars::annual_counts %>%
  filter(source=="FARS", involved != "any") %>%
  ggplot(aes(x=year, y=n)) +
  geom_col() +
  facet_wrap(.~involved, scales = "free_y") +
  labs(title = "Annual fatal crashes by factor involved", subtitle = "Derived from FARS data files", x=NULL, y=NULL) +
  theme_minimal() +
  theme(plot.title.position = "plot")

rfars::annual_counts %>%
  filter(source=="CRSS", involved != "any") %>%
  ggplot(aes(x=year, y=n)) +
  geom_col() +
  facet_wrap(.~involved, scales = "free_y") +
  labs(title = "Annual crashes of all severity levels by factor involved", subtitle = "Derived from CRSS data files", x=NULL, y=NULL) +
  theme_minimal() +
  theme(plot.title.position = "plot")
```

### Generating Custom Counts

We can use get_fars() and then counts() to generate a variety of custom counts.
Below we pull two years of data:

```{r, message=FALSE}
myFARS <- get_fars(years = c(2022, 2023), proceed = T)
```

Then we can use `counts()` to get the monthly number of crashes in Virginia:

```{r, results='asis'}
my_counts <- counts(
  df = myFARS,
  where = list(states = "VA"),
  what = "crashes",
  interval = c("year", "month")
  )
```

This returns the following dataframe:

```{r, results='asis'}
knitr::kable(my_counts, format = "html")
```

Which we can graph:

```{r}
my_counts %>%
  mutate_at("year", factor) %>%
  ggplot(aes(x=month, y=n, group=year, color=year, label=scales::comma(n))) +
  geom_line(linewidth = 1.5) + 
  labs(x=NULL, y=NULL, title = "Fatal Crashes in Virginia") +
  theme_minimal() +
  theme(plot.title.position = "plot")
```

or

```{r}
my_counts %>%
  mutate(date = lubridate::make_date(year, month)) %>%
  ggplot(aes(x=date, y=n, label=scales::comma(n))) +
  geom_col() + 
  labs(x=NULL, y=NULL, title = "Fatal Crashes in Virginia")  +
  theme(plot.title.position = "plot")
```

We could alternatively count annual fatalities:

```{r, results='asis'}
counts(
  myFARS,
  where = list(states = "VA"),
  what = "fatalities",
  interval = c("year")
  ) %>%
  knitr::kable(format = "html")
```

Or fatalities involving speeding:

```{r, results='asis'}
counts(
  df = myFARS,
  where = list(states = "VA"),
  what = "fatalities",
  interval = c("year"),
  involved = "speeding"
) %>%
  knitr::kable(format = "html")
```

Or fatalities involving speeding in rural areas:

```{r, results='asis'}
counts(
  myFARS,
  where = list(states = "VA", urb="rural"),
  what = "fatalities",
  interval = c("year"),
  involved = "speeding"
) %>%
  knitr::kable(format = "html")
```

Or we can use involved = 'each' to see all of the problems in one state:

```{r, results='asis'}
counts(
  df = myFARS,
  where = list(states = "VA"),
  what = "crashes",
  interval = "year",
  involved = "each"
) %>%
  pivot_wider(names_from = "year", values_from = "n") %>%
  arrange(desc(`2023`)) %>%
  knitr::kable(format = "html")
```

### Comparing Counts

We can use `compare_counts()` to quickly produce comparison graphs. Below we
compare speeding-related fatalities in rural and urban areas:

```{r}
compare_counts(
  df = myFARS,
  interval = "year",
  involved = "speeding",
  what = "fatalities",
  where = list(states = "VA", urb="rural"),
  where2 = list(states = "VA", urb="urban")
  ) %>%
  ggplot(aes(x=factor(year), y=n, label=scales::comma(n))) + 
    geom_col() + 
    geom_label(vjust=1.2) +
    facet_wrap(.~urb) +
    labs(x=NULL, y=NULL, title = "Speeding-Related Fatalities in Virginia", fill=NULL) 
```

And here we compare speeding-related crashes to those related to distraction:

```{r}
compare_counts(
  df = myFARS,
  where = list(states = "VA"),
  interval = "year",
  involved = "speeding",
  involved2 = "distracted driver",
  what = "crashes",
  ) %>%
  ggplot(aes(x=factor(year), y=n, label=scales::comma(n))) + 
    geom_col() + 
    geom_label(vjust=1.2) +
    facet_wrap(.~involved) +
    labs(x=NULL, y=NULL, title = "Speeding- and Distraction-Related Crashes in Virginia", fill=NULL)
```