Project 2

Vir Shah

Agenda

Crash Summaries
Distribution Tables (Crashes Per Month, Weekday, and Hour)
Most Common PCF Violations, Types of Collisions, and Parties at Fault
Pedestrian Accidents, Injuries and Deaths
Bicyclist Accidents, Injuries, and Deaths

Crash Summaries

Show code

library(tidyverse)
dat <- read_csv("crashes_2023.csv")

dat |>
  summarise(
    crash_total  = n(),
    victim_total = sum(NUMBER_INJURED),
    fatal_total  = sum(NUMBER_KILLED)
  )

# A tibble: 1 × 3
  crash_total victim_total fatal_total
        <int>        <dbl>       <dbl>
1      161852       224642        3560

Show code

crashes <- dat |> 
  distinct(CASE_ID, .keep_all = TRUE)

Distribution Table: Crashes Per Month

Show code

crashes |>
  count(MONTH, name = "count") |>
  mutate(pct = 100 * count / sum(count)) |>
  arrange(MONTH)

# A tibble: 12 × 3
   MONTH count   pct
   <dbl> <int> <dbl>
 1     1 12766  7.89
 2     2 11793  7.29
 3     3 13120  8.11
 4     4 12987  8.02
 5     5 13418  8.29
 6     6 13256  8.19
 7     7 13634  8.42
 8     8 14715  9.09
 9     9 14153  8.74
10    10 14974  9.25
11    11 13636  8.42
12    12 13400  8.28

Distribution Table: Crashes Per Day of Week

Show code

crashes |>
  count(DAY_OF_WEEK, name = "count") |>
  mutate(pct = 100 * count / sum(count)) |>
  arrange(DAY_OF_WEEK)

# A tibble: 7 × 3
  DAY_OF_WEEK count   pct
        <dbl> <int> <dbl>
1           1 21647  13.4
2           2 23504  14.5
3           3 23845  14.7
4           4 23934  14.8
5           5 25478  15.7
6           6 22996  14.2
7           7 20448  12.6

Distribution Table: Crashes Per Hour

Show code

crashes |>
  count(HOUR, name = "count") |>
  mutate(pct = 100 * count / sum(count)) |>
  arrange(HOUR)

# A tibble: 24 × 3
    HOUR count   pct
   <dbl> <int> <dbl>
 1     0  3552  2.19
 2     1  3193  1.97
 3     2  3031  1.87
 4     3  2124  1.31
 5     4  2215  1.37
 6     5  3279  2.03
 7     6  4468  2.76
 8     7  7211  4.46
 9     8  8420  5.20
10     9  6472  4.00
# ℹ 14 more rows

Top 10 PCF Violations (Most to Least Common)

Show code

crashes |>
  count(PCF_VIOL_CATEGORY, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 10)

# A tibble: 10 × 2
   PCF_VIOL_CATEGORY       count
   <chr>                   <int>
 1 unsafe speed            48018
 2 improper turning        26804
 3 automobile right of way 23465
 4 traffic signs           15197
 5 DUI                     14635
 6 unsafe lane change       6622
 7 unknown                  5114
 8 wrong side of road       3932
 9 pedestrian right of way  3910
10 pedestrian violation     3788

Types of Collision (Most to Least Common)

Show code

crashes |>
  count(TYPE_OF_COLLISION, name = "count") |>
  arrange(desc(count))

# A tibble: 9 × 2
  TYPE_OF_COLLISION  count
  <chr>              <int>
1 rear end           47294
2 broadside          43212
3 hit object         21753
4 sideswipe          18185
5 head-on            10535
6 vehicle/pedestrian 10438
7 overturned          6186
8 other               3592
9 unknown              657

Top 8 Parties at Fault (Most to Least Common)

Show code

dat |>
  count(STWD_VEHTYPE_AT_FAULT, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 8)

# A tibble: 8 × 2
  STWD_VEHTYPE_AT_FAULT                count
  <chr>                                <int>
1 Passenger Car/Station Wagon         107023
2 Not Stated                           18653
3 Pickup or Panel Truck                15787
4 Motorcycle/Scooter                    6470
5 Bicycle                               4288
6 Pedestrian                            3865
7 Truck or Truck Tractor                1684
8 Truck or Truck Tractor with Trailer   1298

Pedestrian Accidents, Injuries, and Deaths

Show code

ped_crashes <- dat |>
  filter(PEDESTRIAN_ACCIDENT == "yes") |>
  distinct(CASE_ID, .keep_all = TRUE)

ped_counts <- ped_crashes |>
  summarise(
    n_ped_accidents = n(),
    n_ped_injured   = sum(COUNT_PED_INJURED),
    n_ped_killed    = sum(COUNT_PED_KILLED)
  )

ped_counts

# A tibble: 1 × 3
  n_ped_accidents n_ped_injured n_ped_killed
            <int>         <dbl>        <dbl>
1           11490         11022          974

Top 5 PCF Violations for Pedestrian Accidents

Show code

ped_crashes |>
  count(PCF_VIOL_CATEGORY, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  PCF_VIOL_CATEGORY       count
  <chr>                   <int>
1 pedestrian right of way  3679
2 pedestrian violation     3654
3 unsafe speed              922
4 unknown                   853
5 improper turning          653

Top 5 Types of Collision for Pedestrian Accidents

Show code

ped_crashes |>
  count(TYPE_OF_COLLISION, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  TYPE_OF_COLLISION  count
  <chr>              <int>
1 vehicle/pedestrian  9897
2 head-on              401
3 broadside            338
4 sideswipe            321
5 rear end             254

Top 5 Party at Fault for Pedestrian Accidents

Show code

ped_crashes |>
  count(STWD_VEHTYPE_AT_FAULT, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  STWD_VEHTYPE_AT_FAULT       count
  <chr>                       <int>
1 Passenger Car/Station Wagon  4603
2 Pedestrian                   3862
3 Not Stated                   2040
4 Pickup or Panel Truck         689
5 Truck or Truck Tractor         77

Bicyclist Accidents, Injuries, and Deaths

Show code

bike_crashes <- dat |>
  filter(BICYCLE_ACCIDENT == "yes") |>
  distinct(CASE_ID, .keep_all = TRUE)

bike_counts <- bike_crashes |>
  summarise(
    n_bike_accidents = n(),
    n_bike_injured   = sum(COUNT_BICYCLIST_INJURED),
    n_bike_killed    = sum(COUNT_BICYCLIST_KILLED)
  )

bike_counts

# A tibble: 1 × 3
  n_bike_accidents n_bike_injured n_bike_killed
             <int>          <dbl>         <dbl>
1             8744           8683           134

Top 5 PCF Violations for Bicycle Accidents

Show code

bike_crashes |>
  count(PCF_VIOL_CATEGORY, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  PCF_VIOL_CATEGORY       count
  <chr>                   <int>
1 automobile right of way  1673
2 wrong side of road       1577
3 improper turning         1325
4 traffic signs            1081
5 unsafe speed             1000

Top 5 Types of Collision for Bicycle Accidents

Show code

bike_crashes |>
  count(TYPE_OF_COLLISION, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  TYPE_OF_COLLISION count
  <chr>             <int>
1 broadside          3896
2 other              1789
3 sideswipe           955
4 head-on             558
5 rear end            524

Top 5 Parties at Fault for Bicycle Accidents

Show code

bike_crashes |>
  count(STWD_VEHTYPE_AT_FAULT, name = "count") |>
  arrange(desc(count)) |>
  slice_head(n = 5)

# A tibble: 5 × 2
  STWD_VEHTYPE_AT_FAULT       count
  <chr>                       <int>
1 Bicycle                      4283
2 Passenger Car/Station Wagon  2399
3 Not Stated                   1372
4 Pickup or Panel Truck         301
5 Other Vehicle                 261

Research Question

Which group was responsible for a higher proportion of crashes they were involved in, pedestrians or bicyclists?

Show code

ped_crashes |>
  summarise(
    prop_ped_at_fault = mean(STWD_VEHTYPE_AT_FAULT == "Pedestrian", na.rm = TRUE))

# A tibble: 1 × 1
  prop_ped_at_fault
              <dbl>
1             0.336

Show code

bike_crashes |>
  summarise(
    prop_bike_at_fault = mean(STWD_VEHTYPE_AT_FAULT == "Bicycle", na.rm = TRUE))

# A tibble: 1 × 1
  prop_bike_at_fault
               <dbl>
1              0.490

We see that pedestrians are at fault for 33.6% of the accidents that they are involved in, while bicyclists are at fault for 49.0% of the accidents that they are involved in. Therefore, bicyclists are more likely than pedestrians to be at fault for an accident that they are involved in.