US Police Fatalities 2015-2020
Hudson Finch-Batista
10/12/2020
Introduction to the Data and Data Cleaning
Import relevant libraries
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(plotly)## Loading required package: ggplot2##
## Attaching package: 'plotly'## The following object is masked from 'package:ggplot2':
##
## last_plot## The following object is masked from 'package:stats':
##
## filter## The following object is masked from 'package:graphics':
##
## layoutlibrary(lubridate)##
## Attaching package: 'lubridate'## The following objects are masked from 'package:base':
##
## date, intersect, setdiff, unionlibrary(ggplot2)
library(RColorBrewer)
library(streamgraph)
library(alluvial)Load the data
setwd("~/Desktop/Python Datasets/hatecrimes")
fatal1 <- read.csv("fatal-police-shootings-data.csv",stringsAsFactors = TRUE)Analyze the structure
str(fatal1)## 'data.frame': 5610 obs. of 17 variables:
## $ id : int 3 4 5 8 9 11 13 15 16 17 ...
## $ name : Factor w/ 5390 levels ""," Austin Wilburly Reid",..: 5017 3323 2634 3599 3779 3096 3080 600 402 3313 ...
## $ date : Factor w/ 1919 levels "2015-01-02","2015-01-03",..: 1 1 2 3 3 3 4 5 5 5 ...
## $ manner_of_death : Factor w/ 2 levels "shot","shot and Tasered": 1 1 2 1 1 1 1 1 1 1 ...
## $ armed : Factor w/ 96 levels "","air conditioner",..: 39 39 88 87 64 39 39 39 88 87 ...
## $ age : int 53 47 23 32 39 18 22 35 34 47 ...
## $ gender : Factor w/ 3 levels "","F","M": 3 3 3 3 3 3 3 3 2 3 ...
## $ race : Factor w/ 7 levels "","A","B","H",..: 2 7 4 7 4 7 4 7 7 3 ...
## $ city : Factor w/ 2532 levels "300 block of State Line Road",..: 2087 37 2477 2018 718 924 386 92 301 1167 ...
## $ state : Factor w/ 51 levels "AK","AL","AR",..: 48 38 17 5 6 37 4 17 13 39 ...
## $ signs_of_mental_illness: Factor w/ 2 levels "False","True": 2 1 1 2 1 1 1 1 1 1 ...
## $ threat_level : Factor w/ 3 levels "attack","other",..: 1 1 2 1 1 1 1 1 2 1 ...
## $ flee : Factor w/ 5 levels "","Car","Foot",..: 4 4 4 4 4 4 2 4 4 4 ...
## $ body_camera : Factor w/ 2 levels "False","True": 1 1 1 1 1 1 1 1 2 1 ...
## $ longitude : num -123.1 -122.9 -97.3 -122.4 -104.7 ...
## $ latitude : num 47.2 45.5 37.7 37.8 40.4 ...
## $ is_geocoding_exact : Factor w/ 2 levels "False","True": 2 2 2 2 2 2 2 2 2 2 ...head(fatal1)## id name date manner_of_death armed age gender race
## 1 3 Tim Elliot 2015-01-02 shot gun 53 M A
## 2 4 Lewis Lee Lembke 2015-01-02 shot gun 47 M W
## 3 5 John Paul Quintero 2015-01-03 shot and Tasered unarmed 23 M H
## 4 8 Matthew Hoffman 2015-01-04 shot toy weapon 32 M W
## 5 9 Michael Rodriguez 2015-01-04 shot nail gun 39 M H
## 6 11 Kenneth Joe Brown 2015-01-04 shot gun 18 M W
## city state signs_of_mental_illness threat_level flee
## 1 Shelton WA True attack Not fleeing
## 2 Aloha OR False attack Not fleeing
## 3 Wichita KS False other Not fleeing
## 4 San Francisco CA True attack Not fleeing
## 5 Evans CO False attack Not fleeing
## 6 Guthrie OK False attack Not fleeing
## body_camera longitude latitude is_geocoding_exact
## 1 False -123.122 47.247 True
## 2 False -122.892 45.487 True
## 3 False -97.281 37.695 True
## 4 False -122.422 37.763 True
## 5 False -104.692 40.384 True
## 6 False -97.423 35.877 TrueClean the data
fatal1$name = as.character(fatal1$name)
fatal1$date = as.character(fatal1$date)
fatal_clean<- fatal1 %>% mutate(date = ymd(date))
head(fatal_clean)## id name date manner_of_death armed age gender race
## 1 3 Tim Elliot 2015-01-02 shot gun 53 M A
## 2 4 Lewis Lee Lembke 2015-01-02 shot gun 47 M W
## 3 5 John Paul Quintero 2015-01-03 shot and Tasered unarmed 23 M H
## 4 8 Matthew Hoffman 2015-01-04 shot toy weapon 32 M W
## 5 9 Michael Rodriguez 2015-01-04 shot nail gun 39 M H
## 6 11 Kenneth Joe Brown 2015-01-04 shot gun 18 M W
## city state signs_of_mental_illness threat_level flee
## 1 Shelton WA True attack Not fleeing
## 2 Aloha OR False attack Not fleeing
## 3 Wichita KS False other Not fleeing
## 4 San Francisco CA True attack Not fleeing
## 5 Evans CO False attack Not fleeing
## 6 Guthrie OK False attack Not fleeing
## body_camera longitude latitude is_geocoding_exact
## 1 False -123.122 47.247 True
## 2 False -122.892 45.487 True
## 3 False -97.281 37.695 True
## 4 False -122.422 37.763 True
## 5 False -104.692 40.384 True
## 6 False -97.423 35.877 TrueCreate variables for year and month
months = vector(mode = "numeric", length = nrow(fatal_clean))
years = vector(mode = "numeric", length = nrow(fatal_clean))
fatal_clean2 <- cbind(fatal_clean,months,years)
head(fatal_clean2)## id name date manner_of_death armed age gender race
## 1 3 Tim Elliot 2015-01-02 shot gun 53 M A
## 2 4 Lewis Lee Lembke 2015-01-02 shot gun 47 M W
## 3 5 John Paul Quintero 2015-01-03 shot and Tasered unarmed 23 M H
## 4 8 Matthew Hoffman 2015-01-04 shot toy weapon 32 M W
## 5 9 Michael Rodriguez 2015-01-04 shot nail gun 39 M H
## 6 11 Kenneth Joe Brown 2015-01-04 shot gun 18 M W
## city state signs_of_mental_illness threat_level flee
## 1 Shelton WA True attack Not fleeing
## 2 Aloha OR False attack Not fleeing
## 3 Wichita KS False other Not fleeing
## 4 San Francisco CA True attack Not fleeing
## 5 Evans CO False attack Not fleeing
## 6 Guthrie OK False attack Not fleeing
## body_camera longitude latitude is_geocoding_exact months years
## 1 False -123.122 47.247 True 0 0
## 2 False -122.892 45.487 True 0 0
## 3 False -97.281 37.695 True 0 0
## 4 False -122.422 37.763 True 0 0
## 5 False -104.692 40.384 True 0 0
## 6 False -97.423 35.877 True 0 0fatal_clean2$months <- as.integer(fatal_clean2$months)
fatal_clean2$years <- as.integer(fatal_clean2$years)
head(fatal_clean2)## id name date manner_of_death armed age gender race
## 1 3 Tim Elliot 2015-01-02 shot gun 53 M A
## 2 4 Lewis Lee Lembke 2015-01-02 shot gun 47 M W
## 3 5 John Paul Quintero 2015-01-03 shot and Tasered unarmed 23 M H
## 4 8 Matthew Hoffman 2015-01-04 shot toy weapon 32 M W
## 5 9 Michael Rodriguez 2015-01-04 shot nail gun 39 M H
## 6 11 Kenneth Joe Brown 2015-01-04 shot gun 18 M W
## city state signs_of_mental_illness threat_level flee
## 1 Shelton WA True attack Not fleeing
## 2 Aloha OR False attack Not fleeing
## 3 Wichita KS False other Not fleeing
## 4 San Francisco CA True attack Not fleeing
## 5 Evans CO False attack Not fleeing
## 6 Guthrie OK False attack Not fleeing
## body_camera longitude latitude is_geocoding_exact months years
## 1 False -123.122 47.247 True 0 0
## 2 False -122.892 45.487 True 0 0
## 3 False -97.281 37.695 True 0 0
## 4 False -122.422 37.763 True 0 0
## 5 False -104.692 40.384 True 0 0
## 6 False -97.423 35.877 True 0 0for (i in 1:nrow(fatal_clean2)){
if (month(fatal_clean2[i,"date"]) == 1){
fatal_clean2[i,"months"] = 1
}else if (month(fatal_clean2[i,"date"]) == 2){
fatal_clean2[i,"months"] = 2
}else if (month(fatal_clean2[i,"date"]) == 3){
fatal_clean2[i,"months"] = 3
}else if (month(fatal_clean2[i,"date"]) == 4){
fatal_clean2[i,"months"] = 4
}else if (month(fatal_clean2[i,"date"]) == 5){
fatal_clean2[i,"months"] = 5
}else if (month(fatal_clean2[i,"date"]) == 6){
fatal_clean2[i,"months"] = 6
}else if (month(fatal_clean2[i,"date"]) == 7){
fatal_clean2[i,"months"] = 7
}else if (month(fatal_clean2[i,"date"]) == 8){
fatal_clean2[i,"months"] = 8
}else if (month(fatal_clean2[i,"date"]) == 9){
fatal_clean2[i,"months"] = 9
}else if (month(fatal_clean2[i,"date"]) == 10){
fatal_clean2[i,"months"] = 10
}else if (month(fatal_clean2[i,"date"]) == 11){
fatal_clean2[i,"months"] = 11
}else if (month(fatal_clean2[i,"date"]) == 12){
fatal_clean2[i,"months"] = 12
}
}
for (i in 1:nrow(fatal_clean2)){
if(year(fatal_clean2[i,"date"]) == 2015){
fatal_clean2[i,"years"] = 2015
}else if (year(fatal_clean2[i,"date"]) == 2016){
fatal_clean2[i,"years"] = 2016
}else if (year(fatal_clean2[i,"date"]) == 2017){
fatal_clean2[i,"years"] = 2017
}else if (year(fatal_clean2[i,"date"]) == 2018){
fatal_clean2[i,"years"] = 2018
}else if (year(fatal_clean2[i,"date"]) == 2019){
fatal_clean2[i,"years"] = 2019
}else if(year(fatal_clean2[i,"date"]) == 2020){
fatal_clean2[i,"years"] = 2020
}
}
fatal_clean2$months <- as.integer(fatal_clean2$months)
fatal_clean2$years <- as.integer(fatal_clean2$years)Data Visualizations
Alluvial and Streamgraphs
Total Number of People Killed per Year by Race
fatal2 <- fatal_clean2 %>% select(years,race) %>% group_by(years,race) %>% count(race)
fatal2## # A tibble: 42 x 3
## # Groups: years, race [42]
## years race n
## <int> <fct> <int>
## 1 2015 "" 29
## 2 2015 "A" 14
## 3 2015 "B" 258
## 4 2015 "H" 172
## 5 2015 "N" 9
## 6 2015 "O" 14
## 7 2015 "W" 498
## 8 2016 "" 56
## 9 2016 "A" 15
## 10 2016 "B" 234
## # … with 32 more rows# Streamgraph
streamgraph(fatal2, key="race", value="n", date="years") %>%
sg_axis_x(1, "year", "%Y")fatal3 <- fatal_clean2 %>% select(race,years) %>% group_by(race,years) %>% count(race) %>% filter(race != "") %>% arrange(years,race)
fatal3## # A tibble: 36 x 3
## # Groups: race, years [36]
## race years n
## <fct> <int> <int>
## 1 A 2015 14
## 2 B 2015 258
## 3 H 2015 172
## 4 N 2015 9
## 5 O 2015 14
## 6 W 2015 498
## 7 A 2016 15
## 8 B 2016 234
## 9 H 2016 160
## 10 N 2016 16
## # … with 26 more rows# Alluvial graph
alluvial_ts(fatal3, wave = .3, ygap = 5, plotdir = 'centred', alpha=.9,
grid = TRUE, grid.lwd = 5, xmargin = 0.1, lab.cex = .7, xlab = 'Year',
ylab = 'Number of People Killed', border = NA, axis.cex = .7, leg.cex = .8,
leg.col='white',
title = "Number of People Killed by Police\n By Year and Race (2015-2020)")
Heatmaps
Mental Illness and Race by Count and Proportion
fatal_clean2 %>% count(race,signs_of_mental_illness) %>% ggplot(mapping = aes(x = race, y=signs_of_mental_illness)) + geom_tile(mapping = aes(fill = n)) + xlab("Race") + ylab("Signs of Mental Illness") + ggtitle("Mental Illness and Race by Count") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
fatal4 <- fatal_clean2 %>% count(race,signs_of_mental_illness) %>% group_by(race) %>% mutate(prop = n/sum(n))
fatal4## # A tibble: 14 x 4
## # Groups: race [7]
## race signs_of_mental_illness n prop
## <fct> <fct> <int> <dbl>
## 1 "" False 488 0.796
## 2 "" True 125 0.204
## 3 "A" False 70 0.753
## 4 "A" True 23 0.247
## 5 "B" False 1132 0.856
## 6 "B" True 190 0.144
## 7 "H" False 764 0.827
## 8 "H" True 160 0.173
## 9 "N" False 64 0.821
## 10 "N" True 14 0.179
## 11 "O" False 37 0.787
## 12 "O" True 10 0.213
## 13 "W" False 1801 0.711
## 14 "W" True 732 0.289fatal4 %>% ggplot(mapping = aes(x = race, y=signs_of_mental_illness)) + geom_tile(mapping = aes(fill = prop))+ xlab("Race") + ylab("Signs of Mental Illness") + ggtitle("Mental Illness and Race by Proportion") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
Threat Level and Race by Count and Proportion
fatal_clean2 %>% count(race,threat_level) %>% ggplot(mapping = aes(x = race, y=threat_level)) + geom_tile(mapping = aes(fill = n))+ xlab("Race") + ylab("Threat Level") + ggtitle("Threat Level and Race by Count") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
fatal5 <- fatal_clean2 %>% count(race,threat_level) %>% group_by(race) %>% mutate(prop = n/sum(n))
fatal5## # A tibble: 20 x 4
## # Groups: race [7]
## race threat_level n prop
## <fct> <fct> <int> <dbl>
## 1 "" attack 407 0.664
## 2 "" other 170 0.277
## 3 "" undetermined 36 0.0587
## 4 "A" attack 51 0.548
## 5 "A" other 41 0.441
## 6 "A" undetermined 1 0.0108
## 7 "B" attack 887 0.671
## 8 "B" other 374 0.283
## 9 "B" undetermined 61 0.0461
## 10 "H" attack 535 0.579
## 11 "H" other 340 0.368
## 12 "H" undetermined 49 0.0530
## 13 "N" attack 44 0.564
## 14 "N" other 30 0.385
## 15 "N" undetermined 4 0.0513
## 16 "O" attack 31 0.660
## 17 "O" other 16 0.340
## 18 "W" attack 1677 0.662
## 19 "W" other 759 0.300
## 20 "W" undetermined 97 0.0383fatal5 %>% ggplot(mapping = aes(x = race, y=threat_level)) + geom_tile(mapping = aes(fill = prop))+ xlab("Race") + ylab("Threat Level") + ggtitle("Threat Level and Race by Proportion") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
Type of Weapon and Race by Count and Proportion
fatal_clean3 <- fatal_clean2 %>% count(armed) %>% filter(n>10)
fatal_clean3## armed n
## 1 212
## 2 ax 24
## 3 baseball bat 18
## 4 box cutter 12
## 5 gun 3183
## 6 gun and car 11
## 7 gun and knife 18
## 8 gun and vehicle 11
## 9 hammer 16
## 10 hatchet 11
## 11 knife 828
## 12 machete 47
## 13 metal pipe 14
## 14 screwdriver 13
## 15 sharp object 13
## 16 sword 23
## 17 Taser 26
## 18 toy weapon 193
## 19 unarmed 356
## 20 undetermined 172
## 21 unknown weapon 79
## 22 vehicle 161armed_types <- fatal_clean3$armed
armed_types## [1] ax baseball bat box cutter
## [5] gun gun and car gun and knife gun and vehicle
## [9] hammer hatchet knife machete
## [13] metal pipe screwdriver sharp object sword
## [17] Taser toy weapon unarmed undetermined
## [21] unknown weapon vehicle
## 96 Levels: air conditioner air pistol Airsoft pistol ax ... wrenchfatal_clean2 %>% filter(armed %in% armed_types) %>% count(race,armed) %>% ggplot(mapping = aes(x = race, y=armed)) + geom_tile(mapping = aes(fill = n))+ xlab("Race") + ylab("Type of Weapon") + ggtitle("Arms and Race") + theme(plot.title = element_text(hjust = 0.5))
fatal6 <- fatal_clean2 %>% count(race,armed) %>% group_by(race) %>% filter(armed %in% armed_types) %>% mutate(prop = n/sum(n))
fatal6## # A tibble: 115 x 4
## # Groups: race [7]
## race armed n prop
## <fct> <fct> <int> <dbl>
## 1 "" "" 9 0.0151
## 2 "" "ax" 3 0.00504
## 3 "" "baseball bat" 2 0.00336
## 4 "" "box cutter" 1 0.00168
## 5 "" "gun" 363 0.610
## 6 "" "gun and car" 2 0.00336
## 7 "" "gun and knife" 3 0.00504
## 8 "" "hammer" 2 0.00336
## 9 "" "knife" 101 0.170
## 10 "" "machete" 6 0.0101
## # … with 105 more rowsfatal6 %>% ggplot(mapping = aes(x = race, y=armed)) + geom_tile(mapping = aes(fill = prop))+ xlab("Race") + ylab("Type of Weapon") + ggtitle("Type of Weapon and Race by Proportion") + theme(plot.title = element_text(hjust = 0.5))
Race and Method of Fleeing by Count and Proportion
fatal_clean2 %>% count(race,flee) %>% ggplot(mapping = aes(x = race, y=flee)) + geom_tile(mapping = aes(fill = n))+ xlab("Race") + ylab("Method of Fleeing") + ggtitle("Race and Fleeing by Count") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
fatal7 <- fatal_clean2 %>% count(race,flee) %>% group_by(race) %>% mutate(prop = n/sum(n))
fatal7## # A tibble: 34 x 4
## # Groups: race [7]
## race flee n prop
## <fct> <fct> <int> <dbl>
## 1 "" "" 52 0.0848
## 2 "" "Car" 90 0.147
## 3 "" "Foot" 59 0.0962
## 4 "" "Not fleeing" 392 0.639
## 5 "" "Other" 20 0.0326
## 6 "A" "" 5 0.0538
## 7 "A" "Car" 6 0.0645
## 8 "A" "Foot" 11 0.118
## 9 "A" "Not fleeing" 71 0.763
## 10 "B" "" 53 0.0401
## # … with 24 more rowsfatal7 %>% ggplot(mapping = aes(x = race, y=flee)) + geom_tile(mapping = aes(fill = prop))+ xlab("Race") + ylab("Method of Fleeing") + ggtitle("Race and Method of Fleeing by Proportion") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
Threat Level and Type of Weapon by Count and Proportion
fatal_clean2 %>% filter(armed %in% armed_types) %>% count(armed,threat_level) %>% ggplot(mapping = aes(x = armed, y=threat_level)) + geom_tile(mapping = aes(fill = n))+coord_flip()+ xlab("Type of Weapon") + ylab("Threat Level") + ggtitle("Threat Level and Type of Weapon by Count") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()## Coordinate system already present. Adding new coordinate system, which will replace the existing one.
fatal8 <- fatal_clean2 %>% count(armed,threat_level) %>% group_by(threat_level) %>% mutate(prop = n/sum(n)) %>% filter(armed %in% armed_types)
fatal8## # A tibble: 53 x 4
## # Groups: threat_level [3]
## armed threat_level n prop
## <fct> <fct> <int> <dbl>
## 1 "" attack 100 0.0275
## 2 "" other 102 0.0590
## 3 "" undetermined 10 0.0403
## 4 "ax" attack 6 0.00165
## 5 "ax" other 18 0.0104
## 6 "baseball bat" attack 9 0.00248
## 7 "baseball bat" other 8 0.00462
## 8 "baseball bat" undetermined 1 0.00403
## 9 "box cutter" attack 7 0.00193
## 10 "box cutter" other 5 0.00289
## # … with 43 more rowsfatal8 %>% ggplot(mapping = aes(x = armed, y=threat_level)) + geom_tile(mapping = aes(fill = prop))+ xlab("Type of Weapon") + ylab("Threat Level") + ggtitle("Threat Level and Type of Weapon by Proportion") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
Threat Level and Signs of Mental Illness by Count and Proportion
fatal_clean2 %>% count(signs_of_mental_illness,threat_level) %>% ggplot(mapping = aes(x = signs_of_mental_illness, y=threat_level)) + geom_tile(mapping = aes(fill = n))+ xlab("Signs of Mental Illness") + ylab("Threat Level") + ggtitle("Threat Level and Signs of Mental Illness by Count") + theme(plot.title = element_text(hjust = 0.5)) 
fatal9 <- fatal_clean2 %>% count(signs_of_mental_illness,threat_level) %>% group_by(threat_level) %>% mutate(prop = n/sum(n))
fatal9## # A tibble: 6 x 4
## # Groups: threat_level [3]
## signs_of_mental_illness threat_level n prop
## <fct> <fct> <int> <dbl>
## 1 False attack 2858 0.787
## 2 False other 1286 0.743
## 3 False undetermined 212 0.855
## 4 True attack 774 0.213
## 5 True other 444 0.257
## 6 True undetermined 36 0.145fatal9 %>% ggplot(mapping = aes(x = signs_of_mental_illness, y=threat_level)) + geom_tile(mapping = aes(fill = prop))+ xlab("Signs of Mental Illness") + ylab("Threat Level") + ggtitle("Threat Level and Signs of Mental Illness by Proportion") + theme(plot.title = element_text(hjust = 0.5)) + coord_flip()
6a. The source of my data was the Washington Post police fatalities dataset. I imported the dataset in such a way that any string variables were regarded as factors since the majority of the string variables appeared to be categorical. There were 17 variables in total, the majority of which were categorical. The date, name, and age variables were the only exceptions. I cleaned the dataset by first converting the name variable to a character variable rather than a factor. I also converted the date variable into a character variable so that I could then easily convert it into a date variable, which I did using the lubridate package. Although I ultimately did not use the date variable directly, I did create two new variables, months and years, which extracted the month and year for each observation in the dataset so that I could group and analyze observations by year or month rather than solely by day. I used for loops to accomplish this although I could’ve also used the case_when function within the mutate function to accomplish this as well. Another cleaning operation I performed consisted in removing categories in the armed variable for which there were less than 10 observations due to the large number of possible values of the armed variable.
6b. The alluvial and streamgraph visualizations at the beginning on my analysis demonstrate that the number of police fatalities by race from 2015 to 2020 has been fairly constant. While it was not particularly surprising that white people constituted the majority of police fatalities from 2015 to 2020 given that the constitute the vast majority of the US population, it was surprising that black people constituted the next largest share of police fatalities from 2015 to 2020 considering that they make up a much smaller percentage of the US population. The same can be said of Hispanics albeit to a lesser degree since they did not die with the same frequency as black people and since they make up a larger proportion of the US population compared to black people. Given the large number of categorical variables, I felt it made sense to looks for associations between categorical variables and so, consequently, I made frequent use of heatmaps. It is interesting to note that for every race, roughly 75% of individuals did not show signs of mental illness, though the percentage seems to be slightly lower for Asian people and white people. It is also interesting that a higher proportion of black and white people were perceived as highly threatening or dangerous as compared to Asians, Hispanics, and Native Americans. The proportion of individuals that were armed with guns was high for every race but Asians, with whites seeming to constitute the highest proportion. Asians did, however, make up the highest proportion of individuals armed with knives, which came as a surprise. Another interesting observation was that the proportion of people that fled from the police was at or above 60% for every race. It would’ve been very interesting to look at age and race or gender and race. I have a hypothesis that the amount of black people killed by police is roughly the same across age and gender, whereas middle-aged white men and Asian men are killed more often than young men or women. I’d like to investigate this conjecture in the future. It would also have been very interesting to analyze the details of the police officers themselves (i.e. the gender, age, race, and number of officers for any given fatality) to see if, for example, black police officers are just as likely to kill black people as white police officers.