Executive Summary

The required libraries were imported into the RStudio interactive development environment, followed by the .csv files “gun-violence-data” and “American Census Data 2015” as gvd and acs2015, respectively 12. These data sets were then checked for tidiness (gvd untidy, acs2015 tidy), dimensions (gvd 239677 obs. of 29 variables, acs2015 3220 obs. of 37 variables), NA count and attribute classes (see summary tables below).
The gvd dataset failed to conform to the “tidy data” principles. Most notably, the attributes prefixed “participant_” contained multiple values per cell, delimited by a combination of [:digit:], :: and/or ||. In order to be computationally efficient, it was decided to first subset gvd before cleaning the aforementioned “participant_” attributes. As the acs2015 dataset contained values from only the 2015 American (US) Census, the first subset involved filtering values of the year 2015 from the gvd dataset.
This was achieved using the as.logical() and year() functions, creating and applying a boolean vector to gvd to create the dataset gvd2015. This methodology enabled us to retain the native date format YYYY-MM-DD. The gvd2015 was then filtered for values which satisfied the condition state == “California”, giving gvd2015_keep_cali. A custom function was written and employed to scour through the participant_age column of gvd2015. This resulted in capturing age counts of incident participants by binning them into approximately 5-years bin intervals. The participant, _gender, _status, _type columns were then also all stripped of string prefix values and binned. The resultant killed_count and injured_count matched perfectly the original n_killed and n_injured attributes, indicating an extremely precise processing methodology. It was determined that the incident_url column of gvd2015_keep_cali could be dropped, as the URL prefix in every case matched the unique incident ID. The acs2015 dataset was then filtered for values from the state of California. Following this step, an inner join was performed between the Gun Violence and American (US) Census datasets.
The joined dataset was checked for NA values, followed by conversion to long format for plotting and value distribution analysis. The data was transformed and checked and cleaned for outliers.

Required packages

#The following packages were used whilst completing the report (with annotation)
library(readr) # Used to import .csv files
library(dplyr) # Section Y
library(stringr) # Section Z
library(knitr) # Section X
library(tidyr)
library(lubridate)
library(kableExtra)
library(outliers)
library(ggplot2)
#xlsx #readxl #foreign #gdata #rvest #dplyr  #deductive #validate #Hmisc 
#MVN #infotheo #MASS #caret #MLR #ggplot #base R functions

Importing the Data

The first dataset, ‘Gun Violence Data - A Comprehensive Record of Over 260k US Gun Violence Incidents from 2013-2018’, was compiled by James Ko and made available on kaggle.com. This dataset contains all recorded gun violence incidents in the US between January 2013 and March 2018 (inclusive).
The second dataset, ‘US Census Demographic Data Demographic and Economic Data for Tracts and Counties’, was collected by the US Census Bureau and made available on kaggle.com. This dataset contains all census data from the 2015 Federal US census.
The readr function was used to import the data as shown below.

gvdZip <- "gun-violence-data.csv.zip"
outDir<-"/Users/alistairgj/Documents/GitHub" # Change output directory for unzipped .csv file as needed
unzip(gvdZip, exdir=outDir)
setwd("/Users/alistairgj/Documents/GitHub")
gvd <- read.csv("gun-violence-data.csv") # Importing Gun Violence Data
setwd("/Users/alistairgj/Documents/GitHub/MATH2349_Assignment3V2")
acs2015 <- read.csv("acs2015_county_data.csv") # Importing American (US) Census Data 2015

Checking the Data

Two custom output table was created to check gvd & acs2015. These results are summarized below.

Feature gvd acs2015
Observations 239677 3220
Attributes 29 37
Tidy No Yes
Format Wide Wide
Unique Ref Yes - IncidentID Yes - CensusID
Incorrect Class Attribute 1, 5, 8:14, 16, 19 Attribute 1
Contains NA Attribute 11, 15, 17, 18, 28, 29 Attribute 14, 15, 19 
data.frame(AttNo = c(1:29), Attribute = names(gvd), Class = sapply(gvd, class),
           gvd_incident486623 = sapply(gvd[84,], function(x) paste0(head(x),  collapse = ", ")),
           NA_Count = sapply(gvd, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% scroll_box(height = "200px")
AttNo Attribute Class gvd_incident486623 NA_Count
1 incident_id integer 486623 0
2 date factor 2013-05-23 0
3 state factor Tennessee 0
4 city_or_county factor Bean Station 0
5 address factor 1034 Main Street 0
6 n_killed integer 2 0
7 n_injured integer 2 0
8 incident_url factor http://www.gunviolencearchive.org/incident/486623 0
9 source_url factor http://legacy.wbir.com/story/news/local/hamblen-jefferson-cocke-greene/2014/10/01/developing-pharmacy-shooting-suspect-to-make-plea-deal/16526831/ 0
10 incident_url_fields_missing factor False 0
11 congressional_district integer 2 11944
12 gun_stolen factor 0::Unknown 0
13 gun_type factor 0::Unknown 0
14 incident_characteristics factor Shot - Wounded/Injured||Shot - Dead (murder, accidental, suicide)||Mass Shooting (4+ victims injured or killed excluding the subject/suspect/perpetrator, one location)||Armed robbery with injury/death and/or evidence of DGU found 0
15 latitude numeric 36.3474 7923
16 location_description factor Down Home Pharmacy 0
17 longitude numeric -83.2849 7923
18 n_guns_involved integer 1 99451
19 notes factor ms; 2 dead; 2 wounded. Pharmacy robbery by ex cop. He was sentenced to life in prison without the possibility of parole. 0
20 participant_age factor 1::72||2::45||4::37 0
21 participant_age_group factor 0::Adult 18+||1::Adult 18+||2::Adult 18+||3::Adult 18+||4::Adult 18+ 0
22 participant_gender factor 0::Male||1::Male||2::Female||3::Female||4::Male 0
23 participant_name factor 0::Stephen Lovell||1::Richard Alexander Sommerville||2::Alexia Gail Wilson||3::Janet Cliff||4::Jason Brian Holt 0
24 participant_relationship factor 4::Armed Robbery 0
25 participant_status factor 0::Killed||1::Killed||2::Injured||3::Injured||4::Unharmed, Arrested 0
26 participant_type factor 0::Victim||1::Victim||2::Victim||3::Victim||4::Subject-Suspect 0
27 sources factor http://www.dailymail.co.uk/news/article-2330162/Former-police-officer-hunt-pain-pills-shoots-dead-pharmacy-owner-customer-robbery.html?ito=feeds-newsxml||http://legacy.wbir.com/story/news/local/hamblen-jefferson-cocke-greene/2014/10/01/developing-pharmacy-shooting-suspect-to-make-plea-deal/16526831/ 0
28 state_house_district integer 35 38772
29 state_senate_district integer 8 32335 
data.frame(AttNo = c(1:37), Attribute = names(acs2015), Class = sapply(acs2015, class),
           acs2015 = sapply(acs2015, function(x) paste(head(x),  collapse = ", ")),
           NA_Count = sapply(acs2015, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),
                                                           font_size = 11) %>% scroll_box(height = "200px")
AttNo Attribute Class acs2015 NA_Count
1 CensusId integer 1001, 1003, 1005, 1007, 1009, 1011 0
2 State factor Alabama, Alabama, Alabama, Alabama, Alabama, Alabama 0
3 County factor Autauga, Baldwin, Barbour, Bibb, Blount, Bullock 0
4 TotalPop integer 55221, 195121, 26932, 22604, 57710, 10678 0
5 Men integer 26745, 95314, 14497, 12073, 28512, 5660 0
6 Women integer 28476, 99807, 12435, 10531, 29198, 5018 0
7 Hispanic numeric 2.6, 4.5, 4.6, 2.2, 8.6, 4.4 0
8 White numeric 75.8, 83.1, 46.2, 74.5, 87.9, 22.2 0
9 Black numeric 18.5, 9.5, 46.7, 21.4, 1.5, 70.7 0
10 Native numeric 0.4, 0.6, 0.2, 0.4, 0.3, 1.2 0
11 Asian numeric 1, 0.7, 0.4, 0.1, 0.1, 0.2 0
12 Pacific numeric 0, 0, 0, 0, 0, 0 0
13 Citizen integer 40725, 147695, 20714, 17495, 42345, 8057 0
14 Income numeric 51281, 50254, 32964, 38678, 45813, 31938 1
15 IncomeErr numeric 2391, 1263, 2973, 3995, 3141, 5884 1
16 IncomePerCap integer 24974, 27317, 16824, 18431, 20532, 17580 0
17 IncomePerCapErr integer 1080, 711, 798, 1618, 708, 2055 0
18 Poverty numeric 12.9, 13.4, 26.7, 16.8, 16.7, 24.6 0
19 ChildPoverty numeric 18.6, 19.2, 45.3, 27.9, 27.2, 38.4 1
20 Professional numeric 33.2, 33.1, 26.8, 21.5, 28.5, 18.8 0
21 Service numeric 17, 17.7, 16.1, 17.9, 14.1, 15 0
22 Office numeric 24.2, 27.1, 23.1, 17.8, 23.9, 19.7 0
23 Construction numeric 8.6, 10.8, 10.8, 19, 13.5, 20.1 0
24 Production numeric 17.1, 11.2, 23.1, 23.7, 19.9, 26.4 0
25 Drive numeric 87.5, 84.7, 83.8, 83.2, 84.9, 74.9 0
26 Carpool numeric 8.8, 8.8, 10.9, 13.5, 11.2, 14.9 0
27 Transit numeric 0.1, 0.1, 0.4, 0.5, 0.4, 0.7 0
28 Walk numeric 0.5, 1, 1.8, 0.6, 0.9, 5 0
29 OtherTransp numeric 1.3, 1.4, 1.5, 1.5, 0.4, 1.7 0
30 WorkAtHome numeric 1.8, 3.9, 1.6, 0.7, 2.3, 2.8 0
31 MeanCommute numeric 26.5, 26.4, 24.1, 28.8, 34.9, 27.5 0
32 Employed integer 23986, 85953, 8597, 8294, 22189, 3865 0
33 PrivateWork numeric 73.6, 81.5, 71.8, 76.8, 82, 79.5 0
34 PublicWork numeric 20.9, 12.3, 20.8, 16.1, 13.5, 15.1 0
35 SelfEmployed numeric 5.5, 5.8, 7.3, 6.7, 4.2, 5.4 0
36 FamilyWork numeric 0, 0.4, 0.1, 0.4, 0.4, 0 0
37 Unemployment numeric 7.6, 7.5, 17.6, 8.3, 7.7, 18 0


The length() & unique() functions were used to determine if gvd$incident_id and acs2015$CensusId contained only unique values. In each case the number of values was equal to the number of observations. Thus incident_id and CensusId may be used as observation identifiers if required.

length(unique(gvd$incident_id)) # Checking the number of unique values for incident_id
## [1] 239677
length(unique(acs2015$CensusId)) # Checking the number of unique values for CensusId
## [1] 3220

Based on the output of the above gvd summary table (data.frame(...), the participant_age and participant_gender were explored to better understand their values. It was noted that these cells store multiple, discrete instances of data. The [:digit:] at the start of each value is presumably used to track the values accross from one participant_ attribute to the next. The first observation below contains only one age value, with four participant values. One avenue for addressing these inconsistencies could involve performing counts for each delimited value. However this is beyond the scope of the current analysis and so this particulat inconsistency will be ignored.

gvd %>% select(participant_age, participant_gender) %>% head(n=8) %>% 
  kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11)
participant_age participant_gender
0::20 0::Male||1::Male||3::Male||4::Female
0::20 0::Male
0::25||1::31||2::33||3::34||4::33 0::Male||1::Male||2::Male||3::Male||4::Male
0::29||1::33||2::56||3::33 0::Female||1::Male||2::Male||3::Male
0::18||1::46||2::14||3::47 0::Female||1::Male||2::Male||3::Female
0::23||1::23||2::33||3::55 0::Female||1::Female||2::Female||3::Female||4::Male||5::Male
0::51||1::40||2::9||3::5||4::2||5::15 0::Male||1::Female||2::Male||3::Female||4::Female||5::Male
0::Male||1::Male||2::Male||3::Male||4::Male

Filtering GVD for Year 2015

The Gun Violence Dataset contains data from 2013 till 2015, while the American Census Data is from 2015. We will therefore create a data subset of the Gun Violence Data from the year 2015 only (as we will later be joining these datasets). The attribute gvd$date was converted to date format (YYYY-MM-DD) using as.Date and checked with str(), below. A Boolean vector was then created for 2015-MM-DD values. This was applied as a mask to gvd, and extracted as the subset gvd2015, which was then checked.

gvd$date <- as.Date(gvd$date) # Converting the gvd$date column and applying to the gvd dataset
str(gvd$date) # Checking the conversion of gvd$date
##  Date[1:239677], format: "2013-01-01" "2013-01-01" "2013-01-01" "2013-01-05" "2013-01-07" ...
year2015_boolean <- as.logical(year(gvd$date) == "2015") #Creating a year 2015 true/false vector
summary(year2015_boolean) # Checking the vector results
##    Mode   FALSE    TRUE 
## logical  186098   53579
gvd2015 <- gvd[year2015_boolean,] # Applying the True/False Vector to filter gvd
summary(gvd2015$date) # Confirming only 2015 data is present
##         Min.      1st Qu.       Median         Mean      3rd Qu. 
## "2015-01-01" "2015-04-14" "2015-07-09" "2015-07-06" "2015-09-29" 
##         Max. 
## "2015-12-31"

The DS gvd2015 is now prepared. As an aside, the first methodology used for extracting YYYY = 2015 value from gvd$date relied on conversion of gvd$date to contain year only (gvd$date <- format.Date(gvd$date, format="%Y")), this was then filtered for values of 2015 (gvd$date_year <- filter(gvd, date_year == "2015")). This methodolgy was abandoned, as it unnecessarily stripped out the month and day data from gvd$date. The applied method is no more complex but retains more information.

Filtering for California

The values of gvd2015$state were aggregated for comparison (below). The output enables us to determine the number of city_or_county in each State, as defined during the composition of the Gun Violence Dataset. For the subsequent analysis, the State of California was selected. This was based on an assumption it would have a large range of discrete and continuous values in the corresponding Census dataset (for example both extremely poor and rich socioeconomic areas). Also note that the methodology outlined below could be applied to any of the states in the gvd dataset, so California will act as a proof of concept.

aggregate(city_or_county ~ state, gvd2015, function(x) length(unique(x))) %>% 
  kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% 
  row_spec(5, bold = T, background = "lightskyblue") %>%
  scroll_box(height = "200px")
state city_or_county
Alabama 212
Alaska 63
Arizona 67
Arkansas 116
California 502
Colorado 105
Connecticut 84
Delaware 36
District of Columbia 4
Florida 374
Georgia 313
Hawaii 28
Idaho 59
Illinois 237
Indiana 164
Iowa 88
Kansas 76
Kentucky 213
Louisiana 167
Maine 120
Maryland 132
Massachusetts 160
Michigan 230
Minnesota 126
Mississippi 158
Missouri 218
Montana 50
Nebraska 39
Nevada 34
New Hampshire 68
New Jersey 167
New Mexico 69
New York 423
North Carolina 348
North Dakota 32
Ohio 306
Oklahoma 147
Oregon 120
Pennsylvania 355
Rhode Island 23
South Carolina 247
South Dakota 35
Tennessee 208
Texas 389
Utah 84
Vermont 61
Virginia 213
Washington 157
West Virginia 119
Wisconsin 196
Wyoming 28 
gvd2015_keep_cali <- filter(gvd2015, state == "California") # Creating the new dataset
aggregate(city_or_county ~ state, gvd2015_keep_cali, function(x) length(unique(x))) %>% 
  kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% 
  row_spec(1, bold = T, background = "lightskyblue") # Check the new dataset
state city_or_county
California 502

GVD 2015 California DS

During data exploration it was determined the following attributes will add no futher value to this analysis and will be dropped.

Attribute Explanation / Justification
address This level of extremely fine detail is not required
source_url This can be found online the other URL incident_url
incident_url_fields_missing This is a boolean attribute which can be checked with filtering
incident_characteristics These strings are extremely varied in content
location_description These strings are extremely varied in content
notes These strings are extremely varied in content & contain no information of value for us
participant_age_group We can determine this from participant_age
participant_name This level of extremely fine detail is not required
sources We have source information from incident_url

The desired columns were selected back into gvd2015_keep_cali, which was then checked in the table below. The attributes congressional_district, state_house_district & state_sentate_district were converted into factors and the dimensions of the new gvd2015_keep_cali were checked.

gvd2015_keep_cali <- select(gvd2015_keep_cali, incident_id, date, state, city_or_county, n_killed, n_injured, incident_url, congressional_district, latitude, longitude, n_guns_involved, participant_age, participant_gender, participant_status, participant_type, state_house_district, state_senate_district)
data.frame(Attribute = names(gvd2015_keep_cali), Class = sapply(gvd2015_keep_cali, class),
           gvd_incident274168 = sapply(gvd2015_keep_cali[53,], function(x) paste0(head(x),  collapse = ", ")),
           NA_Count = sapply(gvd2015_keep_cali, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),
                                                           font_size = 11) %>% scroll_box(height = "200px")
Attribute Class gvd_incident274168 NA_Count
incident_id integer 274386 0
date Date 2015-01-05 0
state factor California 0
city_or_county factor Selma 0
n_killed integer 3 0
n_injured integer 0 0
incident_url factor http://www.gunviolencearchive.org/incident/274386 0
congressional_district integer 21 23
latitude numeric 36.5658 19
longitude numeric -119.621 19
n_guns_involved integer 1 1976
participant_age factor 0::36 0
participant_gender factor 0::Female||1::Female||2::Male 0
participant_status factor 0::Killed||1::Killed||2::Killed 0
participant_type factor 0::Victim||1::Victim||2::Subject-Suspect 0
state_house_district integer 31 24
state_senate_district integer 14 23 
gvd2015_keep_cali$congressional_district <- as.factor(gvd2015_keep_cali$congressional_district)
gvd2015_keep_cali$state_house_district <- as.factor(gvd2015_keep_cali$state_house_district)
gvd2015_keep_cali$state_senate_district <- as.factor(gvd2015_keep_cali$state_senate_district)
dim(gvd2015_keep_cali)
## [1] 3234   17

Incident_URL Attribute

The attribute incident_url was compared with incident_id. Using string manipulation, the suffix http://www.gunviolencearchive.org/incident/ was removed from all values in the incident_url. The resultant _url column and the original _id column were converted into vectors (_id = a, _url = b). The vectors a and b were compared, with a delta value of 0. The _url attribute is therefore deemed unnecessary and will be dropped.

gvd2015_keep_cali$incident_url <-  gvd2015_keep_cali$incident_url %>% 
  str_replace_all("http://www.gunviolencearchive.org/incident/", "") # Cleaving off the suffix 
gvd2015_keep_cali$incident_url <- as.integer(gvd2015_keep_cali$incident_url)
# Create two vectors
a <- as.vector(gvd2015_keep_cali$incident_id)
b <- as.vector(gvd2015_keep_cali$incident_url)
a[!(a %in% b)]
## integer(0)

n_gun_involved Attribute

This attribute contained approximately 60% NA values (Var1(1) = NA) and was dropped.

table(sapply(gvd2015_keep_cali$n_guns_involved, function(y) sum(length(which(is.na(y)))))) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11)
Var1 Freq
0 1258
1 1976

participant_age Attribute

The following function was constructed to iterate through the integer values of the participant_age column and bin the values using a +1 count. The resultant binned counts were spread over 15 new attributes.

gvd2015_keep_cali$matches <- str_match_all(gvd2015_keep_cali$participant_age, "\\d+::(\\d+)")
ages = c(c(0,15)) # First age range is under 15s
for (i in seq(16, 76, 5)) # The other age ranges are every 5 years 
{
  ages <- c(ages, c(i, i+4))
}
ages <- c(ages, c(81,100)) # The last age range is 81 plus
for (i in seq(1,length(ages)-1, 2)) # Each pair of ages in the age ranges
{
  # Construct the column using apply
  # Paste here concatenates the string for the column name
  gvd2015_keep_cali[paste("age", ages[i],ages[i+1], sep="_")] <- apply(gvd2015_keep_cali, 1, function(row) {
    cnt <- 0 # Count of ages in current range starts at 0
    for (x in row$matches) # Iterate through the matches we parsed out
    { if (!grepl(":", x, fixed = TRUE)) # If the match does not contain a : then it is the match group rather than the match
      { age <- as.integer(x) # so convert age as string to integer
          if (age >= ages[i] && age <= ages[i+1]) # and check if age falls within age range for the current column 
        { cnt <- cnt + 1 }
      }
    }
    as.integer(cnt)
  })
}

participant_gender Attribute

The characters ,[:space:], ::, [:digit:] & || were stripped from the participant_gender as seen below. A new char value was subsequently created, participant_gender_unique_check. This value was then checked for unique characters, which yielded only blank (“”), Female and Male. Two str_count functions were then used to generate count columns for gender, these were named male_count and female_count. The column participant_gender will be later dropped.

gvd2015_keep_cali$participant_gender <-  gvd2015_keep_cali$participant_gender %>% 
  str_replace_all(",[:space:]", "") %>% #Remove comma & whitespace
  str_replace_all("::", "") %>%
  str_replace_all("[:digit:]", " ") %>%
  str_replace_all("\\||", "")
participant_gender_unique_check <- as.character(gvd2015_keep_cali[,c("participant_gender")])
ul <- unlist(strsplit(participant_gender_unique_check, " "))
unique(ul)
## [1] ""       "Male"   "Female"
gvd2015_keep_cali$male_count <- str_count(gvd2015_keep_cali$participant_gender, pattern = "Male") # Stringr
gvd2015_keep_cali$female_count <- str_count(gvd2015_keep_cali$participant_gender, pattern = "Female") # Stringr

Participant_status Attribute

The characters ,[:space:], ::, [:digit:] & || were stripped from the participant_status in the same manner as for participant_gender. The same methodology was then used to generate counts (unlist, strsplit, unique) for the participant_status values, giving the following:

participant_status_unique_check <- as.character(gvd2015_keep_cali[,c("participant_status")])
ul <- unlist(strsplit(participant_status_unique_check, " "))
unique(ul)
##  [1] ""                        "Killed"                 
##  [3] "Injured"                 "UnharmedArrested"       
##  [5] "InjuredUnharmedArrested" "Arrested"               
##  [7] "Unharmed"                "InjuredArrested"        
##  [9] "KilledArrested"          "KilledUnharmed"         
## [11] "KilledUnharmedArrested"

Several anomalous values which contain mutually exclusive states of being, were identified as; “InjuredUnharmedArrested”, “KilledInjured”, “KilledArrested”, “KilledUnharmed”, “InjuredUnharmed” and “KilledUnharmedArrested”. The grepl function was then used to create a boolean vector for in each case to search for the source of the value in gvd2015_keep_cali. Once the incident_id was identified, the URL “http://www.gunviolencearchive.org/incident/incident_id” was checked. Each instance was then rationalized and given a more appropriate value using the str_replace_all function.

InjuredUnharmedArrested <- grepl("InjuredUnharmedArrested", gvd2015_keep_cali$participant_status)
gvd2015_keep_cali[InjuredUnharmedArrested,c(1,14)]

The values were all checked as described above and updated as follows:

OriginalValue Explanation / Justification UpdatedValue
InjuredUnharmedArrested This value was created as the reporting of an incident is updated InjuredArrested
KilledInjured This value was created when a person who was initially injured later dies Killed
KilledArrested This value was created when a person who was mortally wounded, arrested and later dies Killed
KilledUnharmed This value was created as the reporting of an incident is updated Killed
InjuredUnharmed This value was created as the reporting of an incident is updated Injured
KilledUnharmedArrested This value was created as the reporting of an incident is updated Killed 
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "InjuredUnharmedArrested", "InjuredArrested")
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "KilledInjured", "Killed")
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "KilledArrested", "Killed")
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "KilledUnharmed", "Killed")
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "InjuredUnharmed", "Injured")
gvd2015_keep_cali$participant_status <- str_replace_all(gvd2015_keep_cali$participant_status, "KilledUnharmedArrested", "Killed")
participant_status_unique_check <- as.character(gvd2015_keep_cali[,c("participant_status")])
ul <- unlist(strsplit(participant_status_unique_check, " "))
unique(ul)
## [1] ""                 "Killed"           "Injured"         
## [4] "UnharmedArrested" "InjuredArrested"  "Arrested"        
## [7] "Unharmed"
gvd2015_keep_cali$injured_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "Injured")
gvd2015_keep_cali$unharmed_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "Unharmed")
gvd2015_keep_cali$killed_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "Killed")
gvd2015_keep_cali$unharmedArrested_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "UnharmedArrested")
gvd2015_keep_cali$arrested_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "Arrested")
gvd2015_keep_cali$injuredArrested_count <- str_count(gvd2015_keep_cali$participant_status, pattern = "InjuredArrested")

Participant_Type Attribute

The characters ,[:space:], ::, [:digit:] & || were stripped from the participant_type in the same manner as for participant_gender. The same methodology was then used to generate counts (unlist, strsplit, unique) for the participant_type values, giving the following;

participant_type_unique_check <- as.character(gvd2015_keep_cali[,c("participant_type")])
ul <- unlist(strsplit(participant_type_unique_check, " "))
unique(ul)
## [1] ""                "Victim"          "Subject-Suspect"
gvd2015_keep_cali$victim_count <- str_count(gvd2015_keep_cali$participant_type, pattern = "Victim")
gvd2015_keep_cali$subjectSuspect_count <- str_count(gvd2015_keep_cali$participant_type, pattern = "Subject-Suspect")

The n_killed & n_injured Attributes

The native n_killed attribute was compared with the generated killed_count attribute - these columns were found to be equal. The same was true for the native n_injured and generated injured_count. The n_killed and n_injured columns were both dropped.

a <- as.vector(gvd2015_keep_cali$n_killed) # Creating a vector
b <- as.vector(gvd2015_keep_cali$killed_count) # Creating a vector
a[!(a %in% b)] # Comparing (differential)
## integer(0)
a <- as.vector(gvd2015_keep_cali$n_injured) # Creating a vector
b <- as.vector(gvd2015_keep_cali$injured_count) # Creating a vector
a[!(a %in% b)] # Comparing (differential)
## integer(0)

Checking and Subsetting

The gvd2015_keep_cali dataset was checked for attribute names and filtered for the the desired attributes to create gvd2015_pt2.

colnames(gvd2015_keep_cali)
##  [1] "incident_id"            "date"                  
##  [3] "state"                  "city_or_county"        
##  [5] "n_killed"               "n_injured"             
##  [7] "incident_url"           "congressional_district"
##  [9] "latitude"               "longitude"             
## [11] "n_guns_involved"        "participant_age"       
## [13] "participant_gender"     "participant_status"    
## [15] "participant_type"       "state_house_district"  
## [17] "state_senate_district"  "matches"               
## [19] "age_0_15"               "age_16_20"             
## [21] "age_21_25"              "age_26_30"             
## [23] "age_31_35"              "age_36_40"             
## [25] "age_41_45"              "age_46_50"             
## [27] "age_51_55"              "age_56_60"             
## [29] "age_61_65"              "age_66_70"             
## [31] "age_71_75"              "age_76_80"             
## [33] "age_81_100"             "male_count"            
## [35] "female_count"           "injured_count"         
## [37] "unharmed_count"         "killed_count"          
## [39] "unharmedArrested_count" "arrested_count"        
## [41] "injuredArrested_count"  "victim_count"          
## [43] "subjectSuspect_count"
gvd2015_pt2 <- select(gvd2015_keep_cali, incident_id, date, state, city_or_county, congressional_district, state_house_district, state_senate_district, latitude, longitude, age_0_15, age_16_20, age_21_25, age_26_30, age_31_35, age_36_40, age_41_45, age_46_50, age_51_55, age_56_60, age_61_65, age_66_70, age_71_75, age_76_80, age_81_100, male_count, female_count, injured_count, unharmed_count, killed_count, unharmedArrested_count, arrested_count, injuredArrested_count, victim_count, subjectSuspect_count)
data.frame(AttNo = c(1:34), Attribute = names(gvd2015_pt2), Class = sapply(gvd2015_pt2, class),
           gvd2015_pt2 = sapply(gvd2015_pt2, function(x) paste(head(x),  collapse = ", ")),
           NA_Count = sapply(gvd2015_pt2, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11)
AttNo Attribute Class gvd2015_pt2 NA_Count
1 incident_id integer 271979, 272551, 272555, 271970, 271975, 275619 0
2 date Date 2015-01-01, 2015-01-01, 2015-01-01, 2015-01-01, 2015-01-01, 2015-01-01 0
3 state factor California, California, California, California, California, California 0
4 city_or_county factor Grand Terrace, Merced, Seaside, Downey, San Jose, Bermuda Dunes 0
5 congressional_district factor 31, 16, 20, 40, 19, 36 23
6 state_house_district factor 47, 21, 29, 58, 27, 56 24
7 state_senate_district factor 20, 12, 17, 32, 15, 28 23
8 latitude numeric 34.0322, 37.3006, 36.6121, 33.9327, 37.3339, 33.7429 19
9 longitude numeric -117.326, -120.498, -121.836, -118.102, -121.855, -116.274 19
10 age_0_15 integer 0, 0, 0, 0, 0, 0 0
11 age_16_20 integer 0, 0, 0, 0, 0, 0 0
12 age_21_25 integer 1, 0, 0, 0, 0, 0 0
13 age_26_30 integer 1, 0, 0, 0, 0, 0 0
14 age_31_35 integer 0, 0, 0, 0, 0, 0 0
15 age_36_40 integer 0, 0, 0, 0, 0, 0 0
16 age_41_45 integer 0, 0, 0, 0, 0, 0 0
17 age_46_50 integer 1, 0, 0, 0, 0, 0 0
18 age_51_55 integer 0, 1, 0, 0, 0, 0 0
19 age_56_60 integer 0, 0, 0, 0, 0, 0 0
20 age_61_65 integer 0, 0, 0, 0, 0, 0 0
21 age_66_70 integer 0, 0, 0, 0, 0, 0 0
22 age_71_75 integer 0, 0, 0, 0, 0, 2 0
23 age_76_80 integer 0, 0, 0, 0, 0, 0 0
24 age_81_100 integer 0, 0, 0, 0, 0, 0 0
25 male_count integer 2, 1, 0, 0, 2, 1 0
26 female_count integer 0, 0, 0, 0, 0, 1 0
27 injured_count integer 2, 0, 0, 0, 2, 1 0
28 unharmed_count integer 0, 0, 0, 0, 0, 0 0
29 killed_count integer 1, 1, 0, 0, 0, 1 0
30 unharmedArrested_count integer 0, 0, 0, 0, 0, 0 0
31 arrested_count integer 0, 0, 0, 0, 0, 0 0
32 injuredArrested_count integer 0, 0, 0, 0, 0, 0 0
33 victim_count integer 3, 1, 0, 0, 2, 1 0
34 subjectSuspect_count integer 0, 0, 0, 0, 0, 1 0

American Census Data

The previously imported American (US) Census Data was in the table summarized below.

data.frame(AttNo = c(1:37), Attribute = names(acs2015), Class = sapply(acs2015, class),
           acs2015 = sapply(acs2015, function(x) paste(head(x),  collapse = ", ")),
           NA_Count = sapply(acs2015, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) #%>% scroll_box(height = "200px")
AttNo Attribute Class acs2015 NA_Count
1 CensusId integer 1001, 1003, 1005, 1007, 1009, 1011 0
2 State factor Alabama, Alabama, Alabama, Alabama, Alabama, Alabama 0
3 County factor Autauga, Baldwin, Barbour, Bibb, Blount, Bullock 0
4 TotalPop integer 55221, 195121, 26932, 22604, 57710, 10678 0
5 Men integer 26745, 95314, 14497, 12073, 28512, 5660 0
6 Women integer 28476, 99807, 12435, 10531, 29198, 5018 0
7 Hispanic numeric 2.6, 4.5, 4.6, 2.2, 8.6, 4.4 0
8 White numeric 75.8, 83.1, 46.2, 74.5, 87.9, 22.2 0
9 Black numeric 18.5, 9.5, 46.7, 21.4, 1.5, 70.7 0
10 Native numeric 0.4, 0.6, 0.2, 0.4, 0.3, 1.2 0
11 Asian numeric 1, 0.7, 0.4, 0.1, 0.1, 0.2 0
12 Pacific numeric 0, 0, 0, 0, 0, 0 0
13 Citizen integer 40725, 147695, 20714, 17495, 42345, 8057 0
14 Income numeric 51281, 50254, 32964, 38678, 45813, 31938 1
15 IncomeErr numeric 2391, 1263, 2973, 3995, 3141, 5884 1
16 IncomePerCap integer 24974, 27317, 16824, 18431, 20532, 17580 0
17 IncomePerCapErr integer 1080, 711, 798, 1618, 708, 2055 0
18 Poverty numeric 12.9, 13.4, 26.7, 16.8, 16.7, 24.6 0
19 ChildPoverty numeric 18.6, 19.2, 45.3, 27.9, 27.2, 38.4 1
20 Professional numeric 33.2, 33.1, 26.8, 21.5, 28.5, 18.8 0
21 Service numeric 17, 17.7, 16.1, 17.9, 14.1, 15 0
22 Office numeric 24.2, 27.1, 23.1, 17.8, 23.9, 19.7 0
23 Construction numeric 8.6, 10.8, 10.8, 19, 13.5, 20.1 0
24 Production numeric 17.1, 11.2, 23.1, 23.7, 19.9, 26.4 0
25 Drive numeric 87.5, 84.7, 83.8, 83.2, 84.9, 74.9 0
26 Carpool numeric 8.8, 8.8, 10.9, 13.5, 11.2, 14.9 0
27 Transit numeric 0.1, 0.1, 0.4, 0.5, 0.4, 0.7 0
28 Walk numeric 0.5, 1, 1.8, 0.6, 0.9, 5 0
29 OtherTransp numeric 1.3, 1.4, 1.5, 1.5, 0.4, 1.7 0
30 WorkAtHome numeric 1.8, 3.9, 1.6, 0.7, 2.3, 2.8 0
31 MeanCommute numeric 26.5, 26.4, 24.1, 28.8, 34.9, 27.5 0
32 Employed integer 23986, 85953, 8597, 8294, 22189, 3865 0
33 PrivateWork numeric 73.6, 81.5, 71.8, 76.8, 82, 79.5 0
34 PublicWork numeric 20.9, 12.3, 20.8, 16.1, 13.5, 15.1 0
35 SelfEmployed numeric 5.5, 5.8, 7.3, 6.7, 4.2, 5.4 0
36 FamilyWork numeric 0, 0.4, 0.1, 0.4, 0.4, 0 0
37 Unemployment numeric 7.6, 7.5, 17.6, 8.3, 7.7, 18 0

Subsetting Census Data

For the following analysis, only economic and population data were carried over. The acs2015 dataset was subset into acs2015_keep, below. The subset data contained only 1 NA value. As the GVD dataset was filtered for the State of California, so to will be the acs2015_keep dataset. The California county count was noted to be 58 (as opposed to 502 for GVD). This was retionalized by the data collection process, which was initially generated in police reports. The classification system for county areas used by law enforcement is highly unlikely to match what is used to collect census data. We can consider county classification in the GVD to perhaps be more generalized, whereas in the American Census Data classification to be more empirical.

acs2015_keep <- select(acs2015, State, County, TotalPop, Men, Women, Income, IncomePerCap, Poverty, Unemployment)
data.frame(AttNo = c(1:9), Attribute = names(acs2015_keep), Class = sapply(acs2015_keep, class),
           acs2015_keep = sapply(acs2015_keep, function(x) paste(head(x),  collapse = ", ")),
           NA_Count = sapply(acs2015_keep, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) #%>% scroll_box(height = "200px")
AttNo Attribute Class acs2015_keep NA_Count
1 State factor Alabama, Alabama, Alabama, Alabama, Alabama, Alabama 0
2 County factor Autauga, Baldwin, Barbour, Bibb, Blount, Bullock 0
3 TotalPop integer 55221, 195121, 26932, 22604, 57710, 10678 0
4 Men integer 26745, 95314, 14497, 12073, 28512, 5660 0
5 Women integer 28476, 99807, 12435, 10531, 29198, 5018 0
6 Income numeric 51281, 50254, 32964, 38678, 45813, 31938 1
7 IncomePerCap integer 24974, 27317, 16824, 18431, 20532, 17580 0
8 Poverty numeric 12.9, 13.4, 26.7, 16.8, 16.7, 24.6 0
9 Unemployment numeric 7.6, 7.5, 17.6, 8.3, 7.7, 18 0 
aggregate(County ~ State, acs2015_keep, function(x) length(unique(x))) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% row_spec(5, bold = T, background = "lightskyblue") %>% scroll_box(height = "200px")
State County
Alabama 67
Alaska 29
Arizona 15
Arkansas 75
California 58
Colorado 64
Connecticut 8
Delaware 3
District of Columbia 1
Florida 67
Georgia 159
Hawaii 5
Idaho 44
Illinois 102
Indiana 92
Iowa 99
Kansas 105
Kentucky 120
Louisiana 64
Maine 16
Maryland 24
Massachusetts 14
Michigan 83
Minnesota 87
Mississippi 82
Missouri 115
Montana 56
Nebraska 93
Nevada 17
New Hampshire 10
New Jersey 21
New Mexico 33
New York 62
North Carolina 100
North Dakota 53
Ohio 88
Oklahoma 77
Oregon 36
Pennsylvania 67
Puerto Rico 78
Rhode Island 5
South Carolina 46
South Dakota 66
Tennessee 95
Texas 254
Utah 29
Vermont 14
Virginia 133
Washington 39
West Virginia 55
Wisconsin 72
Wyoming 23 
acs2015_keep_cali <- filter(acs2015_keep, State == "California") # Subsetting acs for California
aggregate(County ~ State, acs2015_keep_cali, function(x) length(unique(x))) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% row_spec(1, bold = T, background = "lightskyblue")
State County
California 58

Joining the Data Sets

The gvd2015_pt2 and acs2015_keep_cali datasets were joined by County. As an inner join was used, as this will retain the County data found only in both sets.

gvd2015_pt2 <- rename(gvd2015_pt2, County = city_or_county) # Renaming 
gvd_acs_join <- inner_join(gvd2015_pt2, acs2015_keep_cali, by = "County")
gvd_acs_join <- subset(gvd_acs_join, select = -c(State, state)) # State attribute dropped (all values = California)
data.frame(Attribute = names(gvd_acs_join), Class = sapply(gvd_acs_join, class),
           gvd_acs_join = sapply(gvd_acs_join, function(x) paste0(head(x),  collapse = ", ")),
           NA_count = sapply(gvd_acs_join, function(y) sum(length(which(is.na(y))))),
           row.names = NULL) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11) %>% scroll_box(height = "200px")
Attribute Class gvd_acs_join NA_count
incident_id integer 272551, 272526, 272165, 273497, 273249, 273495 0
date Date 2015-01-01, 2015-01-01, 2015-01-01, 2015-01-01, 2015-01-02, 2015-01-03 0
County character Merced, Fresno, San Francisco, Merced, Fresno, Fresno 0
congressional_district factor 16, 16, 12, 16, 16, 16 5
state_house_district factor 21, 31, 17, 21, 23, 31 5
state_senate_district factor 12, 14, 11, 12, 8, 14 5
latitude numeric 37.3006, 36.7307, 37.7881, 37.3103, 36.7501, 36.7504 5
longitude numeric -120.498, -119.769, -122.395, -120.481, -119.849, -119.797 5
age_0_15 integer 0, 0, 0, 0, 0, 0 0
age_16_20 integer 0, 2, 0, 0, 0, 0 0
age_21_25 integer 0, 1, 1, 3, 0, 0 0
age_26_30 integer 0, 1, 0, 0, 0, 0 0
age_31_35 integer 0, 0, 0, 0, 1, 0 0
age_36_40 integer 0, 0, 0, 1, 0, 1 0
age_41_45 integer 0, 0, 0, 0, 0, 0 0
age_46_50 integer 0, 0, 0, 0, 0, 0 0
age_51_55 integer 1, 0, 0, 0, 0, 0 0
age_56_60 integer 0, 0, 0, 0, 0, 0 0
age_61_65 integer 0, 0, 0, 0, 0, 0 0
age_66_70 integer 0, 0, 0, 0, 0, 0 0
age_71_75 integer 0, 0, 0, 0, 0, 0 0
age_76_80 integer 0, 0, 0, 0, 0, 0 0
age_81_100 integer 0, 0, 0, 0, 0, 0 0
male_count integer 1, 4, 1, 5, 1, 1 0
female_count integer 0, 0, 0, 0, 0, 0 0
injured_count integer 0, 0, 0, 0, 0, 0 0
unharmed_count integer 0, 4, 1, 5, 1, 0 0
killed_count integer 1, 0, 0, 0, 0, 1 0
unharmedArrested_count integer 0, 4, 1, 5, 1, 0 0
arrested_count integer 0, 4, 1, 5, 1, 0 0
injuredArrested_count integer 0, 0, 0, 0, 0, 0 0
victim_count integer 1, 0, 0, 0, 0, 1 0
subjectSuspect_count integer 0, 4, 1, 5, 1, 0 0
TotalPop integer 263885, 956749, 840763, 263885, 956749, 956749 0
Men integer 133152, 477316, 427909, 133152, 477316, 477316 0
Women integer 130733, 479433, 412854, 130733, 479433, 479433 0
Income numeric 42462, 45233, 81294, 42462, 45233, 45233 0
IncomePerCap integer 18204, 20408, 52220, 18204, 20408, 20408 0
Poverty numeric 26.1, 26.8, 13.2, 26.1, 26.8, 26.8 0
Unemployment numeric 16.2, 13.2, 6.8, 16.2, 13.2, 13.2 0

NA Values in Joined Dataset

The following table shows Counties with corresponding NA values. The number of unique values in all six of the attributes do not match (a match example would be four Counties to 4 congressional_district values). As these columns represent all the of geographical information for each instance, it was determined that the NA values could not be imputed or determined, and were omitted. This reduced our total number of instances by only five.

gvd_acs_join %>% filter(is.na(congressional_district)) %>% 
  select(County, congressional_district, state_house_district, state_senate_district, latitude, longitude) %>%
  kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11)
County congressional_district state_house_district state_senate_district latitude longitude
Fresno NA NA NA NA NA
San Bernardino NA NA NA NA NA
Fresno NA NA NA NA NA
Los Angeles NA NA NA NA NA
Sacramento NA NA NA NA NA 
target <- c("Fresno", "San Bernardino", "Los Angeles", "Sacramento")
gvd_select_county <- filter(gvd_acs_join, County %in% target)
gvd_select_county <- gvd_select_county %>% select(County, congressional_district, state_house_district, state_senate_district, latitude, longitude)
UniqueCount <- lengths(lapply(gvd_select_county, unique))
as.data.frame(UniqueCount) %>% kable() %>% kable_styling(bootstrap_options = c("striped", "hover", "condensed"),  font_size = 11)
UniqueCount
County 4
congressional_district 23
state_house_district 26
state_senate_district 19
latitude 501
longitude 346 
gvd_acs_join <- na.omit(gvd_acs_join)
County congressional_district state_house_district state_senate_district latitude longitude

Wide to Long Format

The age_count section of the dataset was very sparse. For this reason the dataset was gathered accross the age bin values. This dataset was then filtered to remove AgeGroupCounts equal to zero.

gvd_acs_join_gather <- gvd_acs_join %>% gather(age_0_15:age_81_100, key = "AgeGroup", value = "AgeGroupCount")
gvd_acs_age_long <- filter(gvd_acs_join_gather, AgeGroupCount != 0)
bar1 <- ggplot(gvd_acs_age_long, aes(x=AgeGroup)) + 
  geom_bar() +
  ggtitle("Bar Graph of Participant Count versus Age Group") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))
bar1

violin1 <- ggplot(gvd_acs_age_long, aes(x=AgeGroup, y=AgeGroupCount)) + 
  geom_violin() +
  ggtitle("Violin Plot of Age Group Count Density versus Age Group") +
  theme(axis.text.x = element_text(angle = 45, hjust = 1))
violin1

Data Transformation

The latitude and longitude attributes were normalized from 0 till 1 and plotted with an xy scatter.

range01 <- function(x){(x-min(x))/(max(x)-min(x))}
gvd_acs_join$latitude_norm <- range01(gvd_acs_join$latitude)
gvd_acs_join$longitude_norm <- range01(gvd_acs_join$longitude)
scatter3 <- ggplot(gvd_acs_join, aes(latitude_norm, longitude_norm)) + 
  geom_point(aes()) +
  ggtitle("Scatter Plot of Normalized Latitude versus Longitude")
scatter3

Outliers

Outliers were detected using a boxplot of Income per capita.

gvd_acs_join$IncomePerCap %>%  boxplot(main="Box Plot of Income Per Capita", ylab="IncomePerCap", col = "grey")

# Checking the summary stats
summary(gvd_acs_join$IncomePerCap)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17876   20408   27315   27079   28337   52220
# Checking the z score summary stats
z.scores <- gvd_acs_join$IncomePerCap %>% scores(type = "z")
z.scores %>% summary()
##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
## -1.08393 -0.78571  0.02783  0.00000  0.14820  2.96123
# Checking the z score values
gvd_acs_join$IncomePerCap[which(abs(z.scores) > 3)]
## integer(0)

These values could not be classified as outliers according to the z-score method. Outliers were then checked for gvd_acs_join$killed_count. These outliers were also evaluated with the z-score method, and subsequently removed from the dataset.

gvd_acs_join$killed_count %>% boxplot(main="Number Killed in Gun Violence Incident", ylab="Number Killed in Incident", col = "grey")

# Checking the summary stats
summary(gvd_acs_join$killed_count)
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##  0.0000  0.0000  0.0000  0.3438  1.0000 16.0000
# Checking the z score summary stats
z.scores <- gvd_acs_join$killed_count %>% scores(type = "z")
z.scores %>% summary()
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
## -0.4558 -0.4558 -0.4558  0.0000  0.8698 20.7551
# Checking the z score values
gvd_acs_join$killed_count[which(abs(z.scores) > 3)]
## [1]  4  3 16  4
# Imputing the outliers
killed_count_clean <- gvd_acs_join$killed_count[ - which(abs(z.scores) > 3)]
killed_count_clean %>% boxplot(main="Number Killed in Gun Violence Incident", ylab="Number Killed in Incident", col = "grey")

Other

The following scatter plot shows the relationship between income and unemployment, however could not be transformed with log10 as it contains bivariate data.

scatter5 <- ggplot(gvd_acs_join, aes(Income, Unemployment)) + 
  geom_point(aes()) +
  ggtitle("Scatter Plot of Unemployment versus Income")
scatter5

References

  1. https://www.kaggle.com/jameslko/gun-violence-data/data↩

  2. https://www.kaggle.com/zimeiyang/2015-us-census-demographic-data/data↩