In this document, I will be exploring my three chosen datasets on Income Data, Public School Data, and Crash Reporting Data all in Montgomery County with the following information.
Variable Statistics
Missing Information
Strong visualizations
Accompanied by explanations and insights.
Lets First Load in the Datasets
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, union
library(ggplot2)school_data <-read.csv("/Users/xutongzhang/Documents/DATASETS SPRING 2024/Public_Schools_20240211.csv") #Load in School Dataincome_data <-read.csv("/Users/xutongzhang/Documents/DATASETS SPRING 2024/Income Data Final.csv") #Load in Income Datacrash_data <-read.csv("/Users/xutongzhang/Documents/DATASETS SPRING 2024/Crash_Reporting_-_Drivers_Data_20240211.csv") #Load in Crash Data
Descriptive Variable Statistics
This section will show the frequency counts, contingency tables, five number summary, and mean of variables in each of the three datasets.
Public School Data (Montgomery County Maryland)
Categorical Variables: CATEGORY
# Descriptive statistics for ZIP CODEtable(school_data$CATEGORY)
ELEMENTARY SCHOOLS HIGH SCHOOLS MIDDLE SCHOOLS
136 25 39
We can see that there are 136 Elementary Schools, 25 High Schools, and 39 Middle schools. Showing a majority of the schools in Montgomery County are Elementary Schools.
Numerical Variables: ZIP.CODE
summary(school_data$ZIP.CODE)
Min. 1st Qu. Median Mean 3rd Qu. Max.
20814 20853 20876 20871 20901 20912
We notice how zip codes range from 20814 to 20912, a valuable piece of info for connecting our other datasets.
This allows me to count of each type of crash report, such as “Property Damage Crash”, “Injury Crash”, etc. This information will help me understand the distribution of crash severities. We can see how there are 425 fatal crashes, 62,228 crashes resulting in injury, and 10,9452 crashes that resulting in substantial property damage.
Numerical Variables: Speed Limit
# Summary statistics for Speed Limitsummary(crash_data$`Speed.Limit`)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0.00 25.00 35.00 32.55 40.00 75.00
This shows us the five-number summary (minimum, first quartile, median, third quartile, maximum) and mean of the speed limits where crashes occurred, which can help us know whether or not higher speed limits correlate with more severe or frequent crashes.
Contingency Table: Elementary Schools IN ZIP.CODE
# Contingency table for ACRS Report Type and Speed Limittable(crash_data$`ACRS.Report.Type`, crash_data$`Speed.Limit`)
This table will help us see if certain types of crashes are more common at specific speed limits, helping us to see areas where speed limit adjustments might improve safety.
Income Data (Montgomery County Maryland)
Categorical Variables: ZIP Code
# Unique ZIP codeslength(unique(income_data$ZIP_Code))
[1] 478
We see from this is that there are 478 unique Zip codes in the dataset, helping us show the geographic coverage of my income data.
Numerical Variables: Median Household Income and Total Households
# Summary statistics for Median Household Incomesummary(income_data$Median_Household_Income)
Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
19722 77758 100386 106476 129495 246618 56
# Summary statistics for Total Householdssummary(income_data$Total_Households)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0 331 1836 4870 8264 27598
The data on top shows us the five-number summary (minimum, first quartile, median, third quartile, maximum) and mean of the speed limits where crashes occurred, which can help us know whether or not higher speed limits correlate with more severe or frequent crashes.
The data on the bottom provides us with how many households are in each ZIp code from least densely populated to most.
Contingency Table: NA
This dataset does not have sufficient rows and columns to prepare a contingency table. Please stay tuned for the joining with other datasets in the following sections.
Report on Missing Values
In this section, I will be individually be looking at each dataset and reporting the missing values, how many of each, and stating how it will affect my analysis if at all.
School Data Report
sapply(school_data, function(x) sum(is.na(x)))
CATEGORY SCHOOL.NAME ADDRESS CITY ZIP.CODE PHONE
0 0 0 0 0 0
URL LONGITUDE LATITUDE LOCATION
0 0 0 0
The return above shows that there are in fact no missing values across all variables in the Public School dataset. This completeness ensures that our analysis will be as comprehensive as possible.
The return above shows that there is some missing values, specifically:
Cross Street Name: 8
Vehicle Model: 25
Vehicle Make: 11
Considering the number of values in this Dataset, these numbers are extremley small, and only for 3 out of 43 total variables. Having said this, I believe that there will be little to no affect on my analysis, even more so that these are less significant values for my research.
The return above shows that there is only 1/3 variables that are missing data, specifically:
Median Household Income: 56
This is significant as it is the most important variable in my dataset. This effectively removes 56 zipcodes that I could possibly use, making it harder to do my research if it requires those variables. To counteract this I will need to ensure that I focus on zipcodes that do not rely on this income data.
Compelling Visualizations
I will create visualizations in this section for each dataset that show:
Distribution of variables
Relationships inside of their own datasets
Outliers inside their own datasets
Descriptive Statistics where needed
to create comprehensive and understandable visualizations that aid in answering my research direction.
Public School Visualizations
Bar Graph of School Types
school_counts <-table(school_data$CATEGORY) #Count the number of each type# create a bar plotggplot(data =as.data.frame(school_counts), aes(x = Var1, y = Freq)) +geom_bar(stat ="identity", fill =c("skyblue", "orange", "lightgreen")) +labs(title ="Distribution of School Types in Montgomery County",x ="School Type",y ="Number of Schools") +theme_minimal() +theme(axis.text.x =element_text(angle =45, hjust =1))
The chart shows a distribution of school types across Montgomery County providing a clear picture of the number of Elementary, Middle and High schools.
Map of Schools
library(leaflet) #Use leaflet to make interactive mapscolors <-c("ELEMENTARY SCHOOLS"="skyblue", "MIDDLE SCHOOLS"="orange", "HIGH SCHOOLS"="lightgreen") #Establish pretty colors# Add a color column based on school typeschool_data$color <- colors[school_data$CATEGORY]# Create the mapleaflet(school_data) %>%addTiles() %>%# Add default OpenStreetMap map tilesaddCircleMarkers(~LONGITUDE, ~LATITUDE, color =~color,popup =~as.character(CATEGORY), radius =5)
This plot has all schools on a map, color-coded by school type. This visualization will help get a sense of the layout of schools across the county in certain zip codes.
Car Crash Visualizations
Distribution of Crashes by Severity
# Count occurrences each crash typecrash_severity_counts <-table(crash_data$ACRS.Report.Type)# Convert table to dataframe crash_severity_df <-as.data.frame(crash_severity_counts)names(crash_severity_df) <-c("Crash_Severity", "Number_of_Crashes")# Create the bar chartggplot(crash_severity_df, aes(x = Crash_Severity, y = Number_of_Crashes, fill = Crash_Severity)) +geom_bar(stat ="identity", color ="black", show.legend =FALSE) +scale_fill_manual(values =c("Property Damage Crash"="tomato", "Injury Crash"="lightblue", "Fatal Crash"="red")) +theme_minimal() +labs(title ="Distribution of Crashes by Severity",x ="Crash Severity", y ="Number of Crashes") +theme(axis.text.x =element_text(angle =45, hjust =1))
This plot shows the most common type of crashes and their severity. From the looks of it, not that many of the crashes are fatal, but more than half result in serious injury.
Geographical Map of Crashes (Suspected Injury - Fatal)
# Filter out "NO APPARENT INJURY" records because too many slows down mapfiltered_crash_data <- crash_data %>%filter(Injury.Severity !="NO APPARENT INJURY")# colors for different injury level, excluding "NO APPARENT INJURY"severity_colors <-c("POSSIBLE INJURY"="yellow", "SUSPECTED MINOR INJURY"="orange", "SUSPECTED SERIOUS INJURY"="red", "FATAL INJURY"="darkred")# Add color column based on injury severityfiltered_crash_data$color <- severity_colors[filtered_crash_data$`Injury.Severity`]# Create map without "NO APPARENT INJURY"map <-leaflet(filtered_crash_data) %>%addTiles() %>%# OpenStreetMap map tilesaddCircleMarkers(~Longitude, ~Latitude, color =~color,popup =~paste('Injury Severity:', as.character(`Injury.Severity`),'<br>Weather:', as.character(Weather),'<br>Light:', as.character(Light),'<br>Date:', as.Date(Crash.Date.Time),'<br>Car Condition:',as.character(Vehicle.Damage.Extent),'<br>Final Action:', as.character(Vehicle.Movement)),radius =5, opacity =1, fillOpacity =1)# Add legendmap <- map %>%addLegend("bottomright", colors = severity_colors, labels =names(severity_colors),title ="Injury Severity")# Print the mapmap
This plot has all car crashes with a suspected injury to fatal, showing the date, last action of the driver, light condition, and all color coded based on crash severity.
Income Data Visualization
Distribution of Median Household Income by ZIP Code
ggplot(income_data, aes(x = Median_Household_Income)) +geom_histogram(binwidth =10000, fill ="cornflowerblue", color ="black") +theme_minimal() +labs(title ="Distribution of Median Household Income by ZIP Code",x ="Median Household Income",y ="Frequency")
Warning: Removed 56 rows containing non-finite outside the scale range
(`stat_bin()`).
This histogram shows the distribution of medium household income across different zipcodes.
