Census Data Cleaning and Plotting in R

Using R, I extracted census tracts of Alameda county from 2015-2019 census data and merged them with the latitude and longitude values from a corresponding shape file. Next I created a new column for the predominat race of each census tract. This was determined determined if more than 60% of residents were of that particular racial/ethnic group (Hirsch et al., 2023). Each combonation of race + household income received a unique id, VAL_ID (ex: asian & income above 97309, VAL_ID = 7)

# create a predominant race column RACE_PREDOM
alamedaMerged$RACE_PREDOM <- apply(alamedaMerged[, c("RACE_BLACKNH", "RACE_HISP", "RACE_WHITENH", "RACE_ASIANNH","RACE_OTHERNH", "RACE_MULTIPLENH")], 1, function(row) {
  above60_columns <- names(row[row > 0.6])
  above60_columns_no_na <- na.omit(above60_columns)
  if (length(above60_columns_no_na) > 0) {
    above60_columns_no_na
  } else {
    "RACE_MIXED"
  }
})
# find median of INC_MEDHH
median_INC_MEDHH <- median(alamedaMerged$INC_MEDHH, na.rm = TRUE)
alamedaMerged$IS_INCOME_ABOVE_MEDIAN <- ifelse(alamedaMerged$INC_MEDHH > median_INC_MEDHH, 1, 0)

# assign val
alamedaMerged$VAL <- apply(alamedaMerged, 1, function(row) {
  switch(   
    row$RACE_PREDOM,   
    "RACE_WHITENH"= ifelse(row$IS_INCOME_ABOVE_MEDIAN, 1, 2),   
    "RACE_BLACKNH"= 3,
    "RACE_HISP"= 4,
    "RACE_ASIANNH"= ifelse(row$IS_INCOME_ABOVE_MEDIAN, 5, 6),
    "RACE_MIXED"= ifelse(row$IS_INCOME_ABOVE_MEDIAN, 7, 8),
  )
})

Each census tract was assigned a unique color corresponding to it's VAL_ID calculated from the previous code. I plotted the census tract boundaries on a map to visualize the predominat race and household income of each tract in Alameda County. I compared my map to the 2010 census data map of Oakland (Susaneck, 2024). The results were consistent.

# define colors
colors <- c(
  "VALUE_1" = "#00AFF0",  # white, income > 97309
  "VALUE_2" = "#FFFFFF",  # asian, income > 97309
  "VALUE_3" = "#984EA3",  # mixed, income > 97309
  "VALUE_4" = "#ED151E",  # asian, income < 97309FFFFFF
  "VALUE_5" = "#FFFF33",  # mixed, income < 97309
  "VALUE_6" = "#FCB23E",  # hispanic, income < 97309
  "VALUE_7" = "#000000",  # white, income < 97309
  "VALUE_8" = "#8DC73D"  # black, income < 97309
)

# map colors to data
color_mapping <- setNames(colors[unique(alamedaMerged$VAL)], unique(alamedaMerged$VAL))

print(color_mapping[4])

# assign colors and plot
alamedaMerged$SectionColors <- color_mapping[alamedaMerged$VAL]

plot(alamedaMerged$geometry, col = alamedaMerged$SectionColors)

3D Render of Oakland

I used Cesium Ion's UE5 Plugin with its PhotoRealistic 3D Tileset to render a realistic, immersive 3D model of Oakland. By leveraging Cesium's coordinate system and UE5's blueprints, I mapped the user to a census tract aligned with their selected race and income and spawned them in the neighborhood they would likely be living in. This lead to a stark contrast between a lush, green neighborhood with the sounds of birds chirping or houses next to noisy bart tracks and highways without a park in site.

User Interactions with Information Tokens

Interactive icons present facts about the neighborhood as well as highlighting the community efforts being made on the ground that's not always visible from top-down maps. Users learn about how neighborhoods were physically redlined by highways and bart tracks and the disruptive consequences they had on the community it torn through.