3 Assignment 3: NYPD Shooting Incidents — Cleaning, Insights, and Visualization

4 Overview

In this assignment, you will work with NYPD Shooting Incident Data using the nycOpenData package to practice data cleaning, feature engineering, exploratory insight generation, and data visualization with ggplot2.

Compared to previous assignments, this one is more open-ended: you will make a few analytic decisions (e.g., how to define time-of-day categories and which variables to explore for your second insight). This marks your transition from structured exercises to defensible analytic decision-making.

5 Learning Objectives

By completing this assignment, you will be able to:

Retrieve real-world NYC civic data using nycOpenData
Identify and handle missing data in a defensible way
Engineer new features from dates/times and categorical columns
Generate exploratory insights using grouping and summaries
Communicate findings using well-designed ggplot2 visualizations
Write a short interpretation that connects summaries and plots to your insights

6 Textbook Connection

This assignment builds directly from Chapter 3: Visualizations in Reproducible Research Using R.

Students are encouraged to review the chapter before beginning this assignment, as it provides the conceptual foundation and reproducible workflow demonstrated here.

7 Submission Instructions

Submit one .R script file.

Your script must:

Run from top to bottom without errors
Include your insight statements and written summary as comments where requested
Contain two ggplot2 plots with clear titles and labeled axes
Include at least one faceted plot using facet_wrap()

8 Assignment Tasks

8.1 0. Data Source

You must use the nycOpenData package and the function:

nyc_shooting_incidents()

8.2 1. Cleaning and Feature Engineering

Complete all of the following in your script:

8.2.1 1A. Missing data removal (defensible choice)

Choose one column where it is reasonable to remove missing values.
Compute and report (in a comment) how many values are missing in that column before removal.
Remove rows with missing values in that column.
In a comment, explain why that column was “safe” to use.

8.2.2 1B. Lowercase a column

Choose one column (e.g., borough, perp_race, vic_race) and convert values to lowercase.

8.2.3 1C. Create `time_of_day`

Create a new column called time_of_day with three categories:
- Morning
- Afternoon
- Night
You define the time boundaries.
Briefly note your boundaries in a comment.

8.2.4 1D. Create `days_since`

Create a column called days_since showing the number of days between today’s date and the date of the shooting.
Use as.Date() and/or lubridate as needed.

8.3 2. Insights

Provide the following insights as comments in your script:

One insight about time_of_day
Example format: “Shootings are more frequent during ___ than ___ suggestion…”
One additional insight of your choice
Explore anything you find interesting in the dataset (be creative).

8.4 3. Visualizations (ggplot2)

Create two plots using ggplot2:

8.4.1 Plot 3A. `time_of_day` plot

Must directly relate to your time_of_day variable

8.4.2 Plot 3B. personal insight plot

Must support your second insight

Both plots must include:

Color using an aesthetic mapping (aes(color = ...) or aes(fill = ...))
Informative axis labels
A clear title
Custom font/size settings via theme()

Additional requirement:

At least one plot must use facet_wrap()

8.5 4. Written Summary

At the end of your script, write a short paragraph (as comments) describing:

What you discovered from your insights
What the graphs revealed that was not obvious from the raw data

8.6 Starter Code Template

Copy this template into your script and complete the tasks above. (Wickham (2023)) (Spinu, Grolemund, and Wickham (2024)) (Martinez (2026))

#############################################
# Assignment 3: NYPD Shootings
# Cleaning, Insights, and Visualization
#############################################

# ---- Setup ----
library(tidyverse)
library(lubridate)
library(nycOpenData)

# ---- Data Ingest ----
shooting_data <- nyc_shooting_incidents()

# ==============================
# 1. CLEANING & FEATURE ENGINEERING
# ==============================

# 1A. Missingness (choose one column to drop NAs)
# missing_n <- sum(is.na(shooting_data$YOUR_COLUMN))
# Write a comment: how many were missing + why this column is safe
# cleaned <- shooting_data %>% drop_na(YOUR_COLUMN)

# 1B. Lowercase one categorical column
# cleaned <- cleaned %>% mutate(YOUR_COLUMN = str_to_lower(YOUR_COLUMN))

# 1C. Create time_of_day (Morning / Afternoon / Night)
# cleaned <- cleaned %>%
#   mutate(
#     occur_date = as.Date(occur_date),
#     occur_time = hm(occur_time),
#     hour = hour(occur_time),
#     time_of_day = case_when(
#       hour >= ___ & hour < ___ ~ "Morning",
#       hour >= ___ & hour < ___ ~ "Afternoon",
#       TRUE ~ "Night"
#     )
#   )

# 1D. Create days_since
# cleaned <- cleaned %>%
#   mutate(days_since = as.integer(Sys.Date() - as.Date(occur_date)))


# ==============================
# 2. INSIGHTS (write as comments)
# ==============================


# ==============================
# 3. VISUALIZATIONS
# ==============================

# Plot A: time_of_day plot (include facet_wrap on at least one plot)

# Plot B: personal insight plot


# ==============================
# 4. WRITTEN SUMMARY (comments)
# ==============================

9 Reproducibility Practice

This assignment focuses on reproducibility when working with real civic data and date/time feature engineering.

In your script:

Clearly document your analytic choices (especially your time_of_day boundaries) in comments.
Keep your pipeline modular by creating a cleaned object (e.g.,cleaned), and build from it.
After creating your new variables, include a quick checkpoint such as:
- count(cleaned, time_of_day) or
- summary(cleaned$days_since) to verify your features were created correctly.

The goal is that another reader can see exactly what choices you made, reproduce your engineered variables, and understand how you arrived at your two insights.