Hello and welcome to our Introduction to R for Non-Profits Workshop :)

Goals of this R Markdown (Rmd)

1. Import data using an R package connected to an API (Applicatoin Programming Interface)
  - In other words, we don't have to:
   - manually go to a website 
   - hit download
   - move data from downloads to appropriate folder
   - write code (ex. read_csv()) to import the data into R
- Instead, we [request an API key here](https://api.census.gov/data/key_signup.html)
  - Place our unique api key in the code below
  - Import all the census data we would like with the following function
    - get_acs()
  - Explore which variables exist (depending on survey) with the following function
    - load_variables()
2. Visualize Family Poverty Data in Honolulu County
  - Map
  - Time Series
  - Scatter plot with high school diploma attainment

Load in Packages

It may be helpful to think of packages as toolboxes which contain functions, the tools

To save time, these packages are wrapped in a function that installs any package that you do not currently have installed

If you ever receive the error message: “package name” does not exist, the fix is simple: - Go to the console in R Studio and type install.packages(“name of package you want to install”)

knitr::opts_chunk$set(echo = TRUE)
if (!require(librarian)){
  install.packages("librarian")
  library(librarian)
}

librarian::shelf(tidycensus, # star of the show, what we will use to import data
                tidyverse, # most popular R Package for manipulating data
                reactable, # makes fun, nice looking, interactive tables
                plotly, # makes static graphs interactive
                ggiraph, # makes static maps interactive
                ggpubr) # does some statistics and prints the results on our graphs
options(progress_enabled=FALSE) # picky but i didn't want the progress bar appearing in pdf

Define API Key

Request Here Place the API Key emailed to you in the code below replacing 06f9dcf5172cd1b403f9a6c34beea0d7929604f3 with your own

census_api_key("06f9dcf5172cd1b403f9a6c34beea0d7929604f3")

Now that we have access to census data, what variables can we explore?

This code chunk below shows us all the variables from the American Community Survey (ACS) 1 Year Survey Data Profiles for 2022

Data Profiles have the most frequently requested social, economic, housing, and demographic data. Each of these four subject areas is a separate data profile. The Data Profiles summarize the data for a single geographic area, both numbers and percent, to cover the most basic data on all topics.

Here is an amazing guide to navigating census data and using tidycensus

acs_2022_variables <- load_variables(2022, "acs1/profile")

The dataframe we just created (acs_2022_variables) has all the info we need, but is pretty bland Let’s create a table that is prettier and easier to navigate - works best when knitted

reactable(acs_2022_variables, filterable = TRUE, showPageSizeOptions = TRUE, minRows = 10)

Import Data

Let’s investigate family poverty on Oahu at the census tract level

DP03_0119P is the code for Percent!!PERCENTAGE OF FAMILIES AND PEOPLE WHOSE INCOME IN THE PAST 12 MONTHS IS BELOW THE POVERTY LEVEL!!All families

oahu_family_poverty <- get_acs(
  geography = "tract",
  variables = c(percent_of_families_with_income_below_poverty_line = "DP03_0119P"),
  state = "HI",
  county = "Honolulu",
  geometry = TRUE,
  output = "wide",
  year = 2022
) %>% 
  dplyr::filter(GEOID != "15003981200" & GEOID !="15003981900") # Tract in Northwestern Hawaiian Islands that makes map very small and Mamala Bay Golf Course that has %100 percent of families living below the poverty line

Congragulations! You have successfully imported census data into R

Now, let’s map the data

Mapping

family_poverty_map <- ggplot(oahu_family_poverty) +
  geom_sf_interactive(
    aes(
      fill = percent_of_families_with_income_below_poverty_lineE,
      tooltip = paste(NAME, ": ", percent_of_families_with_income_below_poverty_lineE, "%"),
      data_id = NAME
    )
  ) +
  scale_fill_viridis_c(option = "magma") +
  theme_void() 

girafe(ggobj = family_poverty_map, width = 700, height = 400)

Here is a list of the Tract Names with reference numbers

Now let’s look at family poverty in Honolulu County over time

Time Series

years <- 2010:2019
names(years) <- years

family_poverty_2009_2019 <- map_dfr(years, ~{
  get_acs(
    geography = "county",
    variables = c(percent_of_families_with_income_below_poverty_line = "DP03_0119P"),
    state = "HI",
    county = "Honolulu",
    year = .x,
    survey = "acs1"
  )
}, .id = "year")
ggplot(family_poverty_2009_2019, aes(x = year, y = estimate, group = 1)) + 
  geom_line() + 
  geom_point()

This graph is okay, but we can do alot to make it more informative and nicer too look at

ggplot(family_poverty_2009_2019, aes(x = year, y = estimate, group = 1)) + 
  geom_ribbon(aes(ymax = estimate + moe, ymin = estimate - moe), 
              fill = "steelblue",
              alpha = 0.4) + 
  geom_line(color = "steelblue") + 
  geom_point(color = "steelblue", size = 2) + 
  theme_minimal() + 
  #scale_y_continuous(labels = label_dollar(scale = .001, suffix = "k")) + 
  labs(title = "Percentage of Families Under the Poverty Line: Honolulu County",
       x = "Year",
       y = "%",
       caption = "Shaded area represents margin of error around the estimate")

## Scatter Plot

Let’s explore the relationship between poverty and education with a scatter plot

oahu_family_poverty_college_education <- get_acs(
  geography = "tract",
  variables = c(percent_of_families_with_income_below_poverty_line = "DP03_0119P",
                percent_of_indivivuals_with_bachelors_degree_or_higher = "DP02_0065P"),
  state = "HI",
  county = "Honolulu",
  #geometry = TRUE,
  output = "wide",
  year = 2021
) %>% 
  dplyr::filter(GEOID != "15003981200" & GEOID !="15003981900")
ggplot(oahu_family_poverty_college_education, aes(x = percent_of_families_with_income_below_poverty_lineE,
                                             y = percent_of_indivivuals_with_bachelors_degree_or_higherE)) +
  geom_point() +
  geom_smooth() +
  stat_cor() +
  theme_minimal()

This is pretty interesting

With each point being a census tract, our graph hints that there is some relationship between poverty and education at the tract level

    - The number with the "R =" up top tells us that the correlation coefficient is -0.38
      - This is on a scale from -1 to 1
      - Negative correlations describe relationships where one variable increases, the other decreases (and vice versa)
      - Positive correlations describe relationships where variables move together (up or down)
      - Here is a [really neat tool for exploring correlations](https://rpsychologist.com/correlation/)
      
    - So in summary, there is a negative correlation between the variables, but it is not very strong

This may cause you to think!!

    - What variables included in the American Community Survey are the most correlated with family poverty?
        - We can explore this with correlation matricies and machine learning tools that we can discuss in the future

Challenge

Can you replace family poverty with another variable and

  1. Map it
  2. Show the variable over time
  3. Show the variables relationship with education