• I used a US Accident Data Set narrowed down to just Arizona accidents
• The original data set includes 47 columns and approximately 2.8 million records
• Narrowing down to just the Arizona accidents brings that number down to approximately 18,700 records

Link to Dataset: https://www.kaggle.com/datasets/sobhanmoosavi/us-accidents?datasetId=199387&searchQuery=r

• First, I loaded the data. I put the Arizona Accidents file in data and the csv file with information about the Arizona counties in counties

data <- read.csv("AZ_Accidents.csv")
counties <- read.csv("azcounties.csv")

• Next, I imported the following libraries to be used in my project:

library(tidyverse)
library(sf)
library(mapview)
library(ggplot2)
library(dplyr)
library(lubridate)
library(reshape2)

• I changed the Start_Time variable from a chr to a date
• I created a Weekday variable using the new Start_Time variable

data$Start_Time <- mdy(data$Start_Time)
data$Weekday <- lubridate::wday(data$Start_Time, label=T)

data %>%
  select(Severity, Start_Time, Start_Lat, Start_Lng, County, Weather_Condition) %>%
  top_n(5)

## Selecting by Weather_Condition

##   Severity Start_Time Start_Lat Start_Lng   County      Weather_Condition
## 1        2 2016-09-27  33.66886 -112.1284 Maricopa        Widespread Dust
## 2        2 2016-09-27  33.54276 -112.1124 Maricopa        Widespread Dust
## 3        2 2016-08-05  33.38471 -111.9676 Maricopa Thunderstorms and Rain
## 4        2 2016-06-23  33.46216 -112.0236 Maricopa        Widespread Dust
## 5        2 2016-07-18  33.46188 -112.0524 Maricopa Thunderstorms and Rain
## 6        2 2016-07-19  33.29180 -111.9056 Maricopa        Widespread Dust

• Question: Where do the bulk of accidents happen in Arizona? And what could be some of the reasons why they occur in a certain area?
• Other Questions:
• Do certain cities or counties see more crashes than others?
• Is there a time of year or a specific day(s) of the week where more accidents occur?
• Does inclement weather have any affect on amount of severity of crashes?
• Do accidents that are more serious occur on different days than less sever accidents?

data %>%
  group_by(County) %>%
  summarize(Accidents = n()) %>%
  mutate(County = fct_reorder(County, desc(Accidents)))

## # A tibble: 15 × 2
##    County     Accidents
##    <fct>          <int>
##  1 Apache           215
##  2 Cochise          110
##  3 Coconino         563
##  4 Gila             231
##  5 Graham            35
##  6 Greenlee          13
##  7 La Paz           181
##  8 Maricopa       10830
##  9 Mohave           477
## 10 Navajo           222
## 11 Pima            4396
## 12 Pinal            541
## 13 Santa Cruz        33
## 14 Yavapai          766
## 15 Yuma              85

merged <- merge(x=data, y=counties, by="County")

merged %>%
  group_by(County) %>%
  mutate(accidents_per_1000 = (n() / population) * 1000) %>%
  distinct(County, population, accidents_per_1000)

## # A tibble: 15 × 3
## # Groups:   County [15]
##    County     population accidents_per_1000
##    <chr>           <int>              <dbl>
##  1 Apache          66473              3.23 
##  2 Cochise        125092              0.879
##  3 Coconino       144942              3.88 
##  4 Gila            53211              4.34 
##  5 Graham          38145              0.918
##  6 Greenlee         9542              1.36 
##  7 La Paz          16845             10.7  
##  8 Maricopa      4367186              2.48 
##  9 Mohave         211274              2.26 
## 10 Navajo         106609              2.08 
## 11 Pima          1035063              4.25 
## 12 Pinal          420625              1.29 
## 13 Santa Cruz      47463              0.695
## 14 Yavapai        233789              3.28 
## 15 Yuma           202944              0.419

data %>%
  filter(County == "La Paz") %>%
  mapview(xcol = "Start_Lng", ycol = "Start_Lat", zcol = "Severity", crs = 4269, grid = FALSE, label = "Description")

• How many crashes in La Paz County were on the I-10 and how many we not?

data <- data %>%
  mutate(i_10 = case_when(
      str_detect(Street, "I-10") ~ TRUE,
      !str_detect(Street, "I-10") ~ FALSE)) %>%
  mutate(i_10 = as.factor(i_10))
data %>%
  filter(County == "La Paz") %>%
  group_by(i_10) %>%
  summarize(Accidents = n()) %>%
  ggplot(aes(x=i_10, y=Accidents)) + 
  geom_bar(stat="identity", fill="lightblue") +
  labs(x="Accident on the I-10", title="La Paz Accidents (2016-2021)") +
  geom_text(aes(label=Accidents), vjust= 1.5, color="white")

• How many crashes in La Paz County were on the I-10 and how many we not?

• What month did the accidents in La Paz on the I-10 happen in?

data <- data %>%
  mutate(Month = as.factor(month(Start_Time)))
  
data %>%
  filter(County == "La Paz" & i_10 == TRUE) %>%
  group_by(Month) %>%
  summarize(Accidents = n()) %>%
  ggplot(aes(x=Month, y=Accidents)) + 
  geom_bar(stat="identity", fill="lightblue") +
  geom_text(aes(label=Accidents), vjust=1.2, size=2.9, color="white") +
  labs(title="Accidents by Month in La Paz County")

• What month did the accidents in La Paz on the I-10 happen in?

• Of the accidents in La Paz in July, how many of those were on the I-10 and how many were not?

data %>%
  filter(County == "La Paz" & Month == "7") %>%
  group_by(i_10) %>%
  summarize(Accidents = n()) %>%
  ggplot(aes(x=i_10, y=Accidents)) + 
  geom_bar(stat="identity", fill="lightblue") +
  geom_text(aes(label=Accidents), vjust= 1.5, color="white") +
  labs(x="Accident on the I-10", title="La Paz Accidents (2016-2021)")

• Of the accidents in La Paz in July, how many of those were on the I-10 and how many were not?

ggplot(data, aes(factor(Severity), fill=factor(Severity))) + geom_bar(stat = "count") + 
    labs(x = "", y = "Accidents", title = "Accidents by Severity") +
    scale_fill_discrete(name = "Severity")

ggplot(data, aes(factor(Severity), fill=factor(Severity))) + geom_bar(stat = "count") +
  facet_grid(~Sunrise_Sunset) +
  labs(x="Severity", y="Accidents") +
  scale_fill_discrete(name = "Severity")

• What day(s) of the week do more accidents occur, if any?

data_grouped <- data %>%
  group_by(Weekday) %>%
  select(Weekday) %>%
  filter(Weekday != "NA") %>% #filters out about 172 records
  summarize(count = n())

  ggplot(data_grouped, aes(Weekday, count, fill = Weekday)) + 
  geom_bar(stat = "identity", show.legend = FALSE) + 
  labs(x = "", y = "Accidents")

• What day(s) of the week do more accidents occur, if any?

• Do accidents that are more serious occur on different days than less severe accidents?

data %>% 
  group_by(Weekday, Severity) %>%
  summarize(count = n(), .groups = 'keep') %>%
  filter(Weekday != "NA" & Severity > 2) %>%
  ggplot(aes(Weekday, count, fill = Severity)) + 
  geom_bar(stat = "identity", show.legend = FALSE) + 
  facet_grid(~Severity) +
  labs(x = "", y = "Accidents")

• Do accidents that are more serious occur on different days than less severe accidents?

• Does inclement weather have an effect on number of accidents?

data %>%
  group_by(Weather_Condition) %>%
  summarize(count = n()) %>%
  mutate(Weather_Condition = fct_reorder(Weather_Condition, desc(count))) %>%
  top_n(6) %>%
  ggplot(aes(Weather_Condition, count)) + 
  geom_bar(stat = "identity", fill="lightblue") + 
  geom_text(aes(label=count), vjust=1.3, color="white", size=3) +
  labs(x = "", y = "Accidents") + 
  theme(axis.text = element_text(size = 6))

• Does inclement weather have an effect on number of accidents?

data %>% 
  filter(City == "Tempe", Severity > 2) %>%
  mapview(xcol = "Start_Lng", ycol = "Start_Lat", zcol = "Severity", 
          crs = 4269, grid = FALSE, label = "Description")

• La Paz has the highest accidents per 1,000 people with 10.7 and those accidents are presumed to be by travelers between Los Angeles and Phoenix for the most part.
• The accidents in La Paz line up with school breaks and may have a correlation.
• Most accidents have a low severity.
• There is no strong correlation between accident severity and time of day.
• Weekends have far fewer accidents than weekdays. And Friday is the most frequent day for accidents.
• From this data, there is no correlation between bad weather and more car accidents.
• Serious car accidents tend to happen on the highways as opposed to other roads.