DataVisualizationGallery.Rmd

---
title: "Static Data Visualization Gallery"
date: "March 2023"
output: 
  html_document:
    theme: flatly
editor_options: 
  chunk_output_type: console
---

> This is a sample collection of non-interactive data visualizations. See the code used to develop these visualizations on [GitHub](https://github.com/toylyol/misc-data-viz/blob/main/Data%20Viz%20Gallery/DataVisualizationGallery.Rmd). R Markdown was used to create this gallery; the next iteration will, hopefully, be built using Quarto.

```{r setup, include=FALSE}

knitr::opts_chunk$set(eval=TRUE, include=FALSE, echo=FALSE, comment=NA, cache=FALSE)

```

```{r list-development-packages, warnings = FALSE, messages = FALSE, eval = FALSE}

# The following functions are courtesy of Stats and R: https://statsandr.com/blog/an-efficient-way-to-install-and-load-r-packages/

# List names of required packages

packages <- c("tidyverse", "openxlsx", "ggtext", "sf", "ggrepel", "tigris", 
              "rmapshaper", "jsonlite", "sysfonts", "showtext", "here")


# Install required packages if not yet installed
installed_packages <- packages %in% rownames(installed.packages())

if (any(installed_packages == FALSE)) {
  install.packages(packages[!installed_packages])
}


# Load required packages
lapply(packages, library, character.only = TRUE) |> invisible()

```

```{r load-packages, warnings = FALSE, messages = FALSE}

# Load only package actually needed to run this document

library(here)

path <- here()

```

```{r set-up-Google-font, eval = FALSE}

# The Google font used in the flatly theme was identified in this gist: https://gist.github.com/reywood/11069512

# Add Google fonts

sysfonts::font_add_google(name = "Lato",
                          family = "Lato")

showtext::showtext_auto()  # load the font; must be done every session

```


# {.tabset}

## `r fontawesome::fa("chart-line", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

```{r get-slopegraph-data, eval = FALSE}

# Read in cleaned data ----

data <- read.xlsx( paste0(path, "/data/clean_aca_data.xlsx") )


# Restructure data for slopegraph ----

## identify states with the top 10 largest uninsured_rate_change ----

top_states <- data %>% 
  select(state, uninsured_rate_change) %>%
  filter(state != "United States") %>%            # remove aggregate value for entire country so not in top 10
  slice_min(uninsured_rate_change, n = 10) %>%    # replace the superseded top_n() function
  pull(state)                                     # isolate data in one column to save as a character vector to filter later

top_states[[11]] <- "United States" # add aggregate US to the top_states list to use for filtering

## add columns to use in formatting ggplot ----

data <- data %>%
  mutate( pct_change = as.character(round(uninsured_rate_change*100, 0)) ) %>%
  mutate( line_style = ifelse(state == "United States", "dotted", "solid") ) %>%   # add col so USA will have special line type
  mutate( abbrev = case_when(state == "Nevada" ~ "NV",
                             state == "Oregon" ~ "OR",
                             state == "California" ~ "CA",
                             state == "Kentucky" ~ "KY",
                             state == "New Mexico" ~ "NM",
                             state == "West Virginia" ~ "WV",
                             state == "Arkansas" ~ "AR",
                             state == "Florida" ~ "FL",
                             state == "Colorado" ~ "CO",
                             state == "Washington" ~ "WA",
                             state == "United States" ~ "USA") ) %>%
  mutate( largest = case_when(abbrev == "CA" ~ "Yes",                              # add col to use for custom color scale
                              abbrev == "OR" ~ "Yes",
                              abbrev == "NV" ~ "Yes",
                              TRUE ~ "No") )

## subset dataframe ----

df_slopegraph <- data %>% 
  filter(state %in% top_states) %>%
  select(state, uninsured_rate_2010, uninsured_rate_2015, pct_change, line_style, abbrev, largest) %>%
  pivot_longer(!c(state, line_style, abbrev, pct_change, largest),  # delineate cols to keep
               names_to = "date", 
               values_to = "rate") 

df_slopegraph$date <- df_slopegraph$date %>%
  str_replace_all("uninsured_rate_2010", "2010") %>%  # change name
  str_replace_all("uninsured_rate_2015", "2015")

```

```{r generate-slopegraph, eval = FALSE}

# Create slopegraph 

slopegraph <- ggplot(data = df_slopegraph, 
                     aes(x = date, 
                         y = rate, 
                         group = state, 
                         color = largest,                                  # specify col to use to color points and lines
                         linetype = line_style,                            # specify col to change line style for USA
                         label = paste0(abbrev, "  ", rate*100, "%")))  +
  geom_line(alpha = 1, size = 1.25) +
  geom_point(alpha = 1, size = 4) +
  geom_text_repel( data = df_slopegraph %>% filter(date == "2010"),        # label pts directly; format label for 2010
                   family = "Lato",
                   hjust = "left", 
                   fontface = "bold", 
                   size = 3.5, 
                   nudge_x = -.3, 
                   direction = "y") +
  geom_text_repel( data = df_slopegraph %>% filter(date == "2015"),
                   family = "Lato",
                   hjust = "right",
                   fontface = "bold", 
                   size = 3.5, 
                   nudge_x = .3, 
                   direction = "y") +
  scale_color_manual(values = c("Yes" = "#2D708EFF",          # specify custom scale colors using 'largest' col
                                "No" = "#c9c9c9")) +
  scale_linetype_manual(values = c("solid" = "solid",         # specify line types
                                   "dotted" = "dotdash")) +
  scale_x_discrete(expand = c(0, 0)) +                        # remove space between plot and axis
  xlab("\nYear") +                                            # change axes titles, using newline to add space between axes
  ylab("Percentage of Uninsured Persons\n") +
  labs(caption = "Source: Kaggle") +
  theme_minimal() +
  theme(
    text = element_text(family = "Lato",
                        color = "#2C3E50"),   # set the font family for all {ggtext} elements
    plot.title.position = "plot",           # help ensure title aligned with plot
    axis.text.y = element_blank(),          # remove y axis labels
    panel.grid.major.y = element_blank(),   # remove horizontal gridlines 
    panel.grid.minor.y = element_blank(),
    legend.position = "none"                # remove legend
  )

ggsave( here("images", "slopegraph.png"), height = 576, units = "px" )

```

**California, Oregon, and Nevada saw the <span style ='color: #0F4B63'>largest decreases</span> in uninsured persons after the Affordable Care Act.**

On average, the uninsured rate in the United States decreased six percent from 2010 to 2015. The top 10 states with the largest decreases in uninsured persons had decreases of more than eight percent. 

![]( `r here("images/slopegraph.png")` "This is a slopegraph depicting the percentage of uninsured persons in the United States in 2010 and in 2015 (before and after the Affordable Care Act). The states with the top ten deceases in the percentage of uninsured persons are on the y-axis, and the two years (2010 and 2015) are on the x-axis. The average for the United States is also included. The slopegraph shows a decrease in the percentage of uninsured persons from 2010 to 2015.")

## `r fontawesome::fa("info-circle", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

- Thanks to Bruno Kenzo for [inspiration](https://github.com/KenzoBH/Visualizations) on how to display this collection of data visualizations.

- The alt text was generated using the formula provided by Liz Hare in her October 2022 [presentation](https://lizharedogs.github.io/RLadiesNYAltText "Click here to be directed to Liz's presentation slides.") for R-Ladies NYC.

- The data used in this slopegraph was downloaded from [Kaggle](https://www.kaggle.com/datasets/hhs/health-insurance).

- Chuck Powell's [slopegraph tutorial](https://ibecav.github.io/slopegraph/) was invaluable for determining how to structure the data and to use {ggrepel}. 

- This [Stack Overflow forum](https://stackoverflow.com/questions/71573377/cannot-import-fonts-into-r) provided a quick way to load and use Google fonts without downloading them locally.

- A [Thinking on Data](https://www.thinkingondata.com/something-about-viridis-library/) blog post provided the HEX codes in the viridis palette so that a cohesive color palette could be created.

- Many resources were used to remember the options that can be used for {ggplot} customization: 
  - [Removing gridlines](https://r-graphics.org/recipe-appearance-hide-gridlines)
  - [Aligning captions](https://stackoverflow.com/questions/64701500/left-align-ggplot-caption)
  - [Changing color conditionally within ggrepel::geom_text_repel()](https://stackoverflow.com/questions/49622822/control-colour-of-geom-text-repel)
  - [Specifying custom color scale](https://community.rstudio.com/t/setting-colours-in-ggplot-conditional-on-value/8328/2)
  - [Changing line types conditionally](http://www.sthda.com/english/wiki/ggplot2-line-types-how-to-change-line-types-of-a-graph-in-r-software)
  - [Aligning to the plot](https://ggplot2.tidyverse.org/articles/faq-axes.html#how-can-i-remove-the-space-between-the-plot-and-the-axis)

# {-}

<br>


# {.tabset}

## `r fontawesome::fa("map-location-dot", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

```{r get-broadband-data, eval = FALSE}

# Get data ----

## retrieve data from Census Bureau's ACCESS Dashboard

broadband <- openxlsx::read.xlsx( paste0(path,"/data/county_data_ACCESS_BROADBAND_Dashboard.xlsx") )

## retrieve sf object using {tigris}

ut_counties <- tigris::counties(state = "UT", year = 2021) %>%
  rmapshaper::ms_simplify()

## change case

names(ut_counties) <- tolower(names(ut_counties))


# Merge and clean up data ----

ut_counties_sf <- ut_counties %>%
  left_join(broadband, by = c("geoid" = "GEO_ID")) %>%
  select(-c("lsad", "classfp", "mtfcc", "csafp", "cbsafp", "metdivfp", "funcstat")) 


# Create new sf object for counties comprising "Silicon Slopes" ----

## Utah County geoid: "49049"
## Salt Lake County geoid: "49035"
## Summit County geoid: "49043"

## subset counties 

utah_co_sf <- ut_counties_sf[ut_counties_sf$geoid == "49049", ]
salt_lake_co_sf <- ut_counties_sf[ut_counties_sf$geoid == "49035", ]
summit_co_sf <- ut_counties_sf[ut_counties_sf$geoid == "49043", ]

## union counties; use bind_rows() to retain county boundaries

silicon_slopes_sf <- utah_co_sf %>%
  st_union(salt_lake_co_sf) %>%
  st_union(summit_co_sf) %>%
  select(name.x, geometry)

## change name 

silicon_slopes_sf$name.x <- "Silicon Slopes"

```

```{r make-broadband-map, eval = FALSE}

# Plot UT broadband map ----

broadband_map <- ggplot() +
  geom_sf(data = ut_counties_sf,
          aes(fill = pct_telework_ACS17_21),
          color = "white",                                      # change county borders
          size = 0.5) +                                         # change stroke width
  scale_fill_viridis_c(name = "Percent of Teleworking Workers",
                       direction = -1,
                       label = scales::label_number(suffix = "%")) +
  geom_sf(data = silicon_slopes_sf,
          color = "white",
          size = 1.85,
          alpha = 0) +                        # make layer transparent
  geom_sf_text(data = silicon_slopes_sf,
               aes(label = name.x),
               color = "white",
               size = 5,
               face = "bold",
               nudge_x = 1.7,
               nudge_y = -0.2) +
  guides(fill = guide_colorbar(title.position = 'top',   
                               title.hjust = .5, 
                               title.theme = element_text(size = 9),
                               barwidth = unit(12, 'lines'), 
                               barheight = unit(.5, 'lines'))) +  
  labs(caption = "Source: U.S. Census Bureau") +
  theme_void() +
  theme(
    text = element_text(family = "Lato"), 
    plot.caption.position = "plot",                            # move caption to be right-aligned with plot
    legend.position = "top",                                   # move legend above map
    legend.text = element_text(size = 8),                      # size legend title text
    legend.margin = margin(10, 6, 6, 4),                       # add cushion
    plot.margin = margin(10, 0, 0, 0)
  ) 

```

**In 2021, the counties of 'Silicon Slopes' had some the highest percentages of teleworkers in Utah.**

Unsurprisingly, the three counties comprising UT's technology hub (Salt Lake, Summit, and Utah) were among the top five counties with the highest percentages of teleworkers.

![]( `r here("images/broadband_map.png")` "This is a choropleth map of the counties within the state of Utah. The counties are filled based on the percentage of teleworkers in the county in the 2017-2021 five-year American Community Survey estimates. The yellow-green-blue-purple viridis color palette is used such that the higher the percentage of teleworkers, the darker the color of the county. The three counties comprising “Silicon Slopes” are emphasized with a thick, white border to highlight the fact that they have some of the highest percentages of teleworkers in the state.")

## `r fontawesome::fa("info-circle", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

- The telework data were obtained from the U.S. Census Bureau's [ACCESS BROADBAND Dashboard](https://www.census.gov/programs-surveys/community-resilience-estimates/partnerships/ntia/broadband-act.html). This dashboard is a result of work done by the Census Bureau and the National Telecommunications and Information Administration (NTIA) following the passage of the ACCESS BROADBAND Act of 2021. The original source of the teleworker data was the five-year American Community Survey from 2017-2021.

- In the ACCESS Dashboard data dictionary, the definition for the percentage of teleworkers is as follows: "The percentage of workers ages 16 years and older that reported their residential address as the geographic location at which they carried out their occupational activities." See the complete data dictionary online or in "/data/file_layout_ACCESS_BROADBAND_Dashboard.xlsx" within the R project on [GitHub](https://github.com/toylyol/misc-data-viz/tree/main/Data%20Viz%20Gallery/data).

- [*Geomcomputation with R*](https://bookdown.org/robinlovelace/geocompr/geometric-operations.html) was helpful for learning which functions could be used to subset an existing shapefile, and union the resulting shapefiles, to create a new shapefile.

# {-}

<br>


# {.tabset}

## `r fontawesome::fa("table-cells", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

```{r retrieve-heatmap-data, eval = FALSE}

# Load data ----

mcd_data <- read_csv( paste0(path, "/data/mcd_menu.csv") ) # use readr instead of base to avoid substitution of spaces for dots


# Keep only columns with % DV ----

keep_cols <- mcd_data %>%
  select( matches("% Daily Value") ) %>%
  names()

keep_cols <- c("Category", "Item", keep_cols) # add two cols to char vector

mcd_data <- mcd_data %>%
  select( all_of(keep_cols) ) # use select helper function to avoid ambiguous-external-vector warning


# Convert to decimals ---

convertPct <- function(column){column/100}

mcd_data <- mcd_data %>%
  mutate(across(`Total Fat (% Daily Value)`:`Iron (% Daily Value)`,
                convertPct))


# Subset data ---- 

breaky_foods <- c("Egg McMuffin", "Egg White Delight", "Sausage McMuffin", "Steak & Egg McMuffin",
                  "Bacon, Egg & Cheese Biscuit (Regular Biscuit)", "Sausage Biscuit (Regular Biscuit)",
                  "Southern Style Chicken Biscuit (Regular Biscuit)", "Steak & Egg Biscuit (Regular Biscuit)",
                  "Bacon, Egg & Cheese McGriddles", "Sausage McGriddles", "Bacon, Egg & Cheese Bagel",
                  "Steak, Egg & Cheese Bagel", "Big Breakfast (Regular Biscuit)", 
                  "Big Breakfast with Hotcakes (Regular Biscuit)", "Hotcakes",
                  "Cinnamon Melts", "Sausage Burrito","Fruit & Maple Oatmeal")

breaky <- mcd_data %>%
  filter(Item %in% breaky_foods)


# Shape data into long format ----

breaky <- breaky %>%
  pivot_longer(cols = !c(Category, Item),
               names_to = "nutrition",
               values_to = "pct_dv")

breaky$Item <- str_replace_all(breaky$Item, "Regular Biscuit", "Regular")

breaky$nutrition <- str_replace_all(breaky$nutrition, " \\(% Daily Value\\)", "") 

breaky <- breaky %>% 
  mutate(face_format = case_when(Item == "Big Breakfast (Regular)" ~ "bold",
                                 Item == "Big Breakfast with Hotcakes (Regular)" ~ "bold",
                                 TRUE ~ "plain")
  )

```

```{r create-heatmap, eval = FALSE}

# Generate heatmap ----

heatmap <- ggplot(data = breaky, 
                  aes(x = nutrition,
                      y = Item,
                      fill = pct_dv,
                      height = 1,                 # delineate size of tiles
                      width = 1)) +
geom_tile(color = "white",                        # add white border
          size = 0.01) +
geom_text(aes(label = paste0(pct_dv*100, "%")),   # format pct_dv as percent
                color = "white",
                size = 3) +
scale_fill_viridis_c(name = "Nutrition\n(% Daily Value)",
                     direction = -1,
                     labels = scales::percent_format()) +
scale_y_discrete(expand = c(0,0)) +               # remove space between plot and axis
scale_x_discrete(expand = c(0, 0)) +
labs(x = "", 
     y = "",                                      # remove titles from x and y axes
     caption = "Source: Kaggle") +
theme_minimal() +
theme(
      text = element_text(family = "Lato",
                          color = "#2C3E50",
                          size = 12),
      axis.text.x = element_text(angle = -45,     # angle the x-axis text
                                 vjust = -0.01,
                                 hjust = -0.01),
      plot.caption.position = "plot",
      legend.text = element_text(size = 9),       # size legend title text
      plot.margin = margin(10, 0, 0, 0)) + 
coord_fixed()                                     # keep tiles square


ggsave( here("images", "heatmap.png"), width = 889, height = 764, units = "px" )

```

**McDonald's 'Big Breakfast' is a high-cholesterol start to the day.**

Of the breakfast items listed, the Big Breakfast (with or without hotcakes) has the worst nutritional value.

![]( `r here("images/heatmap.png")` "This is a heatmap. On the y-axis, McDonald’s breakfast menu items are listed. On the x-axis, several nutritional categories are listed: calories, carbs, cholesterol, dietary fiber, iron, saturated fact, sodium, total fat, vitamin A, and vitamin C. The yellow-green-blue-purple viridis color palette is used: The higher the percentage (percent daily value) of a particular nutritional category that a breakfast item has, the darker the color of the square. For the vitamins and minerals, most of the squares are yellow or light yellow-green. The squares are darkest for nutritional categories like cholesterol, saturated fat, and sodium.")

## `r fontawesome::fa("info-circle", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

- The data visualized in this heatmap are from [Kaggle](https://www.kaggle.com/code/kathakaliseth/mcdonald-s-menu-comparative-nutrition-values).

- This Data Science Tutorials' [heatmap tutorial](https://www.r-bloggers.com/2022/10/how-to-create-a-heatmap-in-r/) distributed via R-Bloggers was quite helpful for recalling how the data must be structured for a heatmap made with ggplot2::geom_tile().

- The code for a beautifully formatted horizontal colorbar is courtesy of Cédric Scherer's [workshop materials](https://www.cedricscherer.com/slides/OutlierConf2021_ggplot-wizardry.pdf)  from Data Visualization Society's Outlier Conf 2021. 

- Two Stack Overflow forums ([1](https://stackoverflow.com/questions/37443499/how-to-fix-adjust-the-width-of-each-band-in-ggplot-geom-tile?noredirect=1&lq=1) and [2](https://stackoverflow.com/questions/23897175/adjust-ggplot2-geom-tile-height-and-width)) were helpful for learning how to configure the height/weight arguments to ensure the tiles were large enough when the heatmap was saved as a PNG.

# {-}

<br>


# {.tabset}

## `r fontawesome::fa("map", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

```{r get-map-data, eval = FALSE}

# Retrieve subset of Multiple Chronic Conditions (MCC) dataset using CMS API ----
# 
# url <- "https://data.cms.gov/data-api/v1/dataset/15b08729-6ea2-4789-bf1a-b96b1da8338f/data?filter[Bene_Demo_Lvl][value]=Sex&filter[Bene_Geo_Lvl][value]=State"
# 
# mcc_data <-  jsonlite::fromJSON(url)
# 
# write.xlsx( mcc_data, paste0(path, "/data/mcc_data.xlsx") ) # save a copy of the MCC data


# Load MCC data (if you don'tfeel like using the CMS API) ----

mcc_data <- read.xlsx( paste0(path, "/data/mcc_data.xlsx") )


# Ensure correct data type ----

mcc_data <- mcc_data %>% 
  mutate( across(c(Prvlnc, Tot_Mdcr_Stdzd_Pymt_PC, Tot_Mdcr_Pymt_PC, Hosp_Readmsn_Rate, ER_Visits_Per_1000_Benes), 
                 as.numeric) )


# Subset MCC data ----

data <- mcc_data %>%
  filter( Bene_Age_Lvl == "<65",
          Bene_Demo_Desc == "Female",
          Bene_MCC == "6+" )


# Load hexagonal U.S. shapefile ----

shapefile <- paste0(path,"/data/us_states_hexgrid.shp") # download from CARTO

hex_map <- read_sf(shapefile)


# Clean up shapefile columns ----

hex_map <- hex_map %>%
  mutate( google_nam = gsub(" \\(United States\\)", "", google_nam) ) %>%
  rename( state_name = google_nam,
          abbrev = iso3166_2 ) %>%
  select( -c(created_at, updated_at, label, bees) ) # remove necessary columns


# Join (non-spatial) with MCC data ----

hex_map <- hex_map %>%
  left_join( data, by = c("state_name" = "Bene_Geo_Desc") )


# Retrieve centroid for labelling ----

hex_map <-  cbind(hex_map, st_coordinates(st_centroid(hex_map)))

```

```{r make-hexmap, eval = FALSE}

# Create hexbin map ----

hex_map <- hex_map %>%
  ggplot() +
  geom_sf(aes(fill = Prvlnc,                                            # indicate col to make chloropleth
              shape = "No data available"),                             # create an override value for NA values
          color = "white") +                                            # change hexbin border colors
  scale_fill_viridis_c(name = "Prevalence of 6+ Chronic Conditions",    # give the legend a name
                       direction = -1,                                  # reverse the scale, so darker equals larger number
                       labels = scales::percent_format(),               # format the numbers in the legend
                       na.value = "gray68") +                           # specify color NA values
  geom_text(aes(x = X,                                                  # specify long of centroid
                y = Y,                                                  # specify lat of centroid
                label = abbrev),                                        # indicate the col to label each hexbin
            color = "white",                                            # change font color
            family = "Lato") +                                          # delineate the custom font
  guides(shape = guide_legend(override.aes = list(fill = "gray68",      # add NA value to legend
                                                  color = "white"),     # set border w/in legend
                              order = 2,
                              title = NULL),
         fill = guide_colorbar(title.position = 'top',                  # move legend title to top
                               title.hjust = .5,                        # use Cédric Scherer code to alter viridis colorbar
                               title.theme = element_text(size = 9), 
                               barwidth = unit(20, 'lines'), 
                               barheight = unit(.5, 'lines'), 
                               order = 1)) +                            # ensure viridis scale is first
  labs(caption = "Source: Centers for Medicare & Medicaid Services") +
  theme_void() +
  theme(
    text = element_text(family = "Lato",
                        color = "#2C3E50"),
    plot.title.position = "plot",                                           # help ensure title aligned with plot
    plot.caption.position = "plot",                                         # move caption to be right-aligned with plot
    legend.position = "top",                                                # move legend above map
    legend.text = element_text(size = 8),                                   # size legend title text
    legend.margin = margin(10, 6, 6, 4)                                     # add cushion between subtitle, legend, and map
    ) 

ggsave( here("images", "hex_map.png"), height = 576, units = "px" )

```

**In 2018, Oklahoma had the highest prevalence of elderly, female Medicare enrollees with 6+ chronic conditions.**

The prevalence of six or more chronic conditions among Medicare beneficiaries assigned female at birth aged 65 years or older was 20.5% in OK in 2018.

![]( `r here("images/hex_map.png")` "This is a hexbin map of the United States. Each state (and Washington D.C.) is represented as a hexagon. Each hexbin is filled based on the prevalence of Medicare beneficiaries aged 65 years or older assigned female at birth who have six or more chronic conditions. The yellow-green-blue-purple viridis color palette is used such that the higher the prevalence percentage, the darker the color of the hexagon. There are several states for which there are no data: Those hexbins are colored a medium gray. The darkest colors are in states that would be located in the southeastern United States in a traditional map (e.g., Oklahoma, Florida, Louisiana, etc.), indicating a relatively high prevalence of elderly, female Medicare beneficiaries with multiple chronic conditions.")

## `r fontawesome::fa("info-circle", fill = "#2c3e50", fill_opacity =0.75, height = "1.5em")`

- This visualization is a hexbin map of the United States; it features data from the Centers for Medicare and Medicaid Services' (CMS) Multiple Chronic Conditions (MCC) [dataset](https://data.cms.gov/medicare-chronic-conditions/multiple-chronic-conditions). 

- The MCC data were retrieved through the [CMS API](https://data.cms.gov/api-docs). More information about the dataset can be viewed in the [data dictionary](https://data.cms.gov/resources/multiple-chronic-conditions-data-dictionary) and [methodology documentation](https://data.cms.gov/resources/medicare-chronic-conditions-methodology).

- Yan Holtz's excellent [hexbin map tutorial](https://r-graph-gallery.com/328-hexbin-map-of-the-usa.html) on The R Graph Gallery provided the data source for the hexagonal shapefile—as well as helpful hints for manipulating the file.

- The hexagonal shapefile of the United States was obtained from [Carto](https://team.carto.com/u/andrew/tables/andrew.us_states_hexgrid/public/map). It can be downloaded in other formats like GeoJSON.

- Tips and tricks for handling NA values in a continuous choropleth map were learned from R for the Rest of Us' [Mapping with R](https://rfortherestofus.com/courses/mapping/) course taught by Charlie Hadley.

# {-}

<br>


<p style="text-align: center;">Toyin L. Ola</p>