HW #3: Visualizing FEMA NRI x ACS Data

Learning Outcomes

identify which types of visualizations are most appropriate for your data and your audience
prepare (e.g. clean, explore, wrangle) data so that it’s appropriately formatted for building data visualizations
build effective, responsible, accessible, and aesthetically-pleasing, visualizations using the R programming language, and specifically {ggplot2} + ggplot2 extension packages

Description

In class, we’ve been discussing strategies and considerations for choosing the right graphic form to represent your data and convey your intended message. Here, you’ll apply what we’re learning to natural hazards and demographics data, courtesy of the FEMA Resilience Analysis and Planning Tool (RAPT) and the US Census Bureau’s American Community Survey (ACS).

I. Background reading

Refamiliarize yourself with the FEMA’s NRI and learn more about the ACS. (unfold the following note, which is collapsed to save space):

About the data

About FEMA’s National Risk Index (NRI) for Natural Hazards (repeated from HW #2)

FEMA (Federal Emergency Management Agency) is a government agency with a mission of helping people before, during, and after disasters. In 2021, FEMA launched the National Risk Index (NRI), a dataset that “provides information for communities most at risk to 18 different natural hazards. It offers a baseline risk measurement for expected annual loss, social vulnerability and community resilience at the Census tract or county level. The data helps to validate, measure and better understand your community’s natural hazard risk.” Risk is defined as the potential for negative impacts resulting from natural hazards. It’s calculated using the following equation:

\[Risk\:Index = Expected\:Annual\:Loss \times \frac{Social\:Vulnerability}{Community\:Resilience}\]

NRI provides hazard type-specific scores, as well as a composite score, which adds together the risk from all 18 hazard types. A community’s risk score is represented by its percentile ranking among all other communities at the same level for Risk, Expected Annual Loss, Social Vulnerability and Community Resilience – for example, if a given county’s Risk Index percentile for a hazard type is 84.32 then its Risk Index value is greater than 84.32% of all US counties. Each community is also assigned a risk rating, which is a qualitative rating that describes the community in comparison to all other communities at the same level, ranging from “Very Low” to “Very High.” You can learn more in the National Risk Index Data Technical Documentation.

In December 2025, FEMA integrated NRI data into the Resilience Analysis and Planning Tool (RAPT):

Screenshot of the Resilience Analysis and Planning Tool (RAPT) with the National Risk Index Counties data layer.

About the US Census Bureau’s American Community Survey (ACS)

The American Community Survey (ACS) is a nationwide, continuous survey designed to provide communities with reliable and timely social, economic, housing, and demographic data every year. Unlike the Decennial Census (which counts every person in the US every 10 years for the purpose of congressional appointment), the ACS collects detailed information from a small subset of the population (~3.5 million households) at 1- and 5-year intervals. Learn more about the differences between these 1- and 5-year estimates.

II. Update your `eds240-nri-acs-viz` GitHub repo

Make the following modifications to your eds240-nri-acs-vizrepository:

update your README to reflect the new work you’re adding (you can wait until the end of your assignment to complete this, if you prefer)
add a HW3.qmd file with the necessary YAML

III. Import & save ACS data

Let’s download the necessary data using the US Census Bureau’s API via the {tidycensus} package and write a copy to file. We’ll then read that saved copy of our data back in for use.

The following three steps describe the code you see in the following chunk. You may copy this code into the top of your HW3.qmd file, where you’d normally read in your data (remember to load any necessary packages first, though, and clean up the code comments as necessary).

Import ACS data using tidycensus::get_acs(): You’ll need your API key to use {tidycensus} functions (revisit week 3 pre-class prep instructions, if necessary).
Write your data to .csv: It’s always a good idea to write your data (i.e. the race_ethnicity data frame, from above) to file, in case the Census Bureau’s API goes down. If you write your csv file to your data/ folder, it will automatically be ignored by .gitignore.
Read in your saved .csv file.

#....Step 1a: see all available ACS variables + descriptions.....
acs_vars <- tidycensus::load_variables(year = 2023,
                                       dataset = "acs1")

#..............Step 1b: import race & ethnicity data.............
race_ethnicity <- tidycensus::get_acs(
  geography = "county",
  survey = "acs1",
  # NOTE: you may not end up using all these variables
  variables = c("B01003_001", "B02001_002", "B02001_003", 
                "B02001_004", "B02001_005", "B02001_006",
                "B02001_007", "B02001_008", "B03002_012",
                "B03002_002"),
  state = "CA", 
  year = 2023) |>
  # join variable descriptions (so we know what's what!)
  dplyr::left_join(acs_vars, by = dplyr::join_by(variable == name)) 

#.................Step 2: write ACS data to file.................
readr::write_csv(race_ethnicity, here::here("data", "ACS-1yr-2023-county-race-ethnicity.csv"))

#..................Step 3: read in your CSV file.................
race_ethnicity <- readr::read_csv(here::here("data", "ACS-1yr-2023-county-race-ethnicity.csv"))

At this point, you should comment out code from steps 1 & 2 (lines 1-20, above) so that you’re not pinging the API & overwriting your saved copy each time you render HW3.qmd. Comment these lines out, rather that delete them altogether, so that anyone checking out your work understands how you initially retrieved the data. If you have difficulties downloading the data via the API, you can find a copy on Google Drive.

IV. Build your visualization

Create a data viz that helps to answer the question, How does climate hazard risk exposure vary across racial / ethnic groups in California? This will require some data wrangling first (including joining the NRI and ACS data).

REMINDER: before you start wrangling / building your viz, be sure to…

identify / jot down your variables of interest and consider which data types they are
use online tools like from Data to Viz to help determine appropriate graph types, given your variables
roughly sketch out your plots by hand (see why this step matters!)

Your final visualization should:

include the following racial / ethnic groups: White, Black or African American, American Indian and Alaska Native, Asian, Native Hawaiian and Other Pacific Islander, Some Other Race, Two or More Races, Hispanic or Latino
include a title (short, descriptive) & subtitle (describes main takeaway) (see Fundamentals of Data Visualization, Ch 22 for an example), a caption (describes the data source, e.g. “Data: FEMA National Risk Index (2023 Release)”), and alt text (following the formula you recently practiced in lab; use the fig-alt code chunk option to apply your alt text)
consider and implement strategies for highlighting trends / important information (e.g. arranging data, highlighting data, adjusting scales)
use custom colors (if applicable), rather than ggplot defaults
have an updated theme / polished theme

Is your plot looking squished in your rendered .qmd file? Adjust the aspect ratio!

You can adjust the aspect ratio of your plot (i.e. its width to height ratio) so that your data / groups are easy to read. Add the fig-asp option to your code chunk YAML, which makes adjusting aspect ratios for rendered outputs quite easy. Values > 1 make your plot taller and values < 1 make your plot wider.

V. Answer some questions

1. What are your variables of interest and what kinds of data (e.g. numeric, categorical, ordered, etc.) are they (a bullet point list is fine)?
2. How did you decide which type of graphic form was best suited for answering the question? What alternative graphic forms could you have used instead? Why did you settle on this particular graphic form?
3. Summarize your main finding in no more than two sentences.
4. What modifications did you make to this visualization to make it more easily readable?
5. Is there anything you wanted to implement, but didn’t know how? If so, please describe.

Rubric (specifications)

To recieve a “Satisfactory” on HW #3:

Create one visualization that answers the question, “How does climate hazard risk exposure vary across racial / ethnic groups in California?” and fulfills the requirements listed in Part IV.
All code should follow appropriate styling conventions and include clear, informative annotations. We will not enforce any particular code style guide (you may use conventions from prior courses if you prefer; we recommend following these conventions from the Tidyverse style guide if you’re unsure where to start). You don’t need to explain every line, but your annotations should make it easy to follow your thinking and understand important decisions in your code.
Answer all three Part V questions. There are no strict length requirements for free-response questions, however we expect that you answer them thoughtfully and fully.
Ensure that everything described above is included in your neatly-organized and polished HW3.qmd file – this includes appropriate document YAML (at a minimum, title, author, date), clearly labeled sections, appropriately-formatted prose, appropriate code chunk options (e.g. code should render and execute, but warnings and messages should be suppressed, long data frames should not be printed out, etc.)
All your work is housed in your organized and polished eds240-nri-acs-viz GitHub repository. Your README is updated to include any new relevant information. You’ve verified that your new ACS data are saved to your data/ folder, which is ignored by .gitignore.
Render HW3.qmd as a PDF. Double check that all of your outputs, code, and prose are visible.
Upload your PDF to Gradescope by 11:59pm on Wed 02/11/2026.
Double check that you added your GitHub repo URL to this Google Sheet (you should have done this already as part of HW #2)
Submit a Generative AI Statement of Use for HW #3 by 11:59pm on Wed 02/11/2026.