05. Exploratory data visualization

Author

Camille Seaberry

Modified

September 8, 2026

Exploratory data visualization (EDV) is a method of using visualization to learn about your data. It’s how you’ll make the majority of your charts, eventually pulling out one or two to polish (maybe).

I’m saying exploratory data visualization here rather than exploratory data analysis, because we won’t be doing the basic analyses, summary statistics, crappy hypothesis testing, etc. that would normally make up a more well-rounded EDA. Just the data viz.

Wickham, H., Çetinkaya-Rundel, M., & Grolemund, G. (2023). R for data science (2nd ed.). O’Reilly Media, Incorporated. https://r4ds.hadley.nz/

From Wickham et al. (2023), Exploratory Data Analysis chapter:

There is no rule about which questions you should ask to guide your research. However, two types of questions will always be useful for making discoveries within your data. You can loosely word these questions as: - What type of variation occurs within my variables? - What type of covariation occurs between my variables?

The questions I usually ask myself (or my boss asks me) in doing EDA/EDV are some variation on:

We’ll follow many of the steps of the EDA chapter using the acs dataset in the {justviz} package. For simplicity, we’ll focus on Maryland census tracts and just a few variables dealing with housing and income.

Start by just looking at your data. Get a summary of it. Figure out what each variable means and what its scale should be.

library(ggplot2)

# set a default theme---easier to see some examples
theme_set(theme_minimal())

# make histogram default to a white border around bars
update_geom_defaults("histogram", list(color = "white"))

acs_all_tracts <- justviz::acs |>
    dplyr::filter(level == "tract") |>
    dplyr::select(
        county,
        name,
        total_pop,
        total_hh,
        homeownership,
        total_cost_burden,
        renter_cost_burden,
        owner_cost_burden,
        no_vehicle_hh,
        median_hh_income,
        total_vacant_units,
        pop_density
    )

head(acs_all_tracts)
county name total_pop total_hh homeownership total_cost_burden renter_cost_burden owner_cost_burden no_vehicle_hh median_hh_income total_vacant_units pop_density
Allegany County 24001000100 3334 1403 0.78 0.19 0.20 0.19 0.03 63415 0.30 17.74039
Allegany County 24001000200 4030 1343 0.90 0.19 0.29 0.18 0.05 67724 0.14 83.83255
Allegany County 24001000500 2108 786 0.58 0.36 0.62 0.18 0.23 34545 0.31 476.60452
Allegany County 24001000600 2830 1349 0.73 0.19 0.30 0.15 0.10 52250 0.11 1788.34855
Allegany County 24001000700 3293 1391 0.45 0.41 0.51 0.28 0.18 36673 0.22 4622.72767
Allegany County 24001000800 1890 816 0.47 0.44 0.55 0.32 0.21 28056 0.22 1496.26735 
summary(acs_all_tracts)
       county            name        total_pop        total_hh   
 Length   :1460   Length   :1460   Min.   :    2   Min.   :   0  
 N.unique :  24   N.unique :1460   1st Qu.: 2968   1st Qu.:1138  
 N.blank  :   0   N.blank  :   0   Median : 4057   Median :1553  
 Min.nchar:  11   Min.nchar:  11   Mean   : 4251   Mean   :1618  
 Max.nchar:  22   Max.nchar:  11   3rd Qu.: 5370   3rd Qu.:2030  
                                   Max.   :14748   Max.   :5132  
                                                                 
 homeownership    total_cost_burden renter_cost_burden owner_cost_burden
 Min.   :0.0000   Min.   :0.0500    Min.   :0.0000     Min.   :0.0000   
 1st Qu.:0.5100   1st Qu.:0.2100    1st Qu.:0.3100     1st Qu.:0.1600   
 Median :0.7450   Median :0.2800    Median :0.4400     Median :0.2100   
 Mean   :0.6763   Mean   :0.2974    Mean   :0.4316     Mean   :0.2246   
 3rd Qu.:0.8900   3rd Qu.:0.3700    3rd Qu.:0.5600     3rd Qu.:0.2700   
 Max.   :1.0000   Max.   :0.7400    Max.   :1.0000     Max.   :1.0000   
 NAs    :4        NAs    :4         NAs    :7          NAs    :13       
 no_vehicle_hh     median_hh_income total_vacant_units  pop_density       
 Min.   :0.00000   Min.   :  2499   Min.   :0.00000    Min.   :    0.419  
 1st Qu.:0.02000   1st Qu.: 74966   1st Qu.:0.02000    1st Qu.: 1006.389  
 Median :0.05000   Median :103192   Median :0.05000    Median : 3455.259  
 Mean   :0.09591   Mean   :110249   Mean   :0.07119    Mean   : 4958.910  
 3rd Qu.:0.12000   3rd Qu.:137163   3rd Qu.:0.09000    3rd Qu.: 6796.855  
 Max.   :0.81000   Max.   :250001   Max.   :0.90000    Max.   :56280.606  
 NAs    :4         NAs    :8        NAs    :4

Make sure nothing looks weird:

Variation

First a histogram of median household income values:

ggplot(acs_all_tracts, aes(x = median_hh_income)) +
    geom_histogram(color = "white")
Figure 1

There’s a message and a warning: the message suggests being intentional about the number of bins, and the warning calls our attention to missing values in this column.

Use the next few chunks of code to experiment with bin specifications. Does your understanding of the data’s distribution change?

ggplot(acs_all_tracts, aes(x = median_hh_income)) +
    geom_histogram(bins = 50) # bins can be determined by setting the number of bins

ggplot(acs_all_tracts, aes(x = median_hh_income)) +
    geom_histogram(binwidth = 10000) # or by the width of bins, with a scale corresponding to the x-axis

What are some values of bins or binwidth that seem reasonable? At what point do either of them start to obscure data?

As for the missing values:

acs_all_tracts |>
    dplyr::filter(is.na(median_hh_income))
county name total_pop total_hh homeownership total_cost_burden renter_cost_burden owner_cost_burden no_vehicle_hh median_hh_income total_vacant_units pop_density
Anne Arundel County 24003740400 4522 0 NA NA NA NA NA NA NA 6.398086e+03
Anne Arundel County 24003980000 2 0 NA NA NA NA NA NA NA 4.194018e-01
Baltimore County 24005492500 3491 12 1.00 0.67 NA 0.67 0.00 NA 0.00 3.426183e+03
Prince George’s County 24033802901 4059 1645 0.51 0.26 0.25 0.27 0.22 NA 0.04 5.637175e+03
Washington County 24043011000 4951 0 NA NA NA NA NA NA NA 2.897222e+03
Baltimore city 24510100300 2247 0 NA NA NA NA NA NA NA 3.723204e+04
Baltimore city 24510150900 3820 1332 0.59 0.54 0.86 0.32 0.44 NA 0.19 6.757337e+03
Baltimore city 24510271801 2537 1236 0.43 0.50 0.65 0.31 0.59 NA 0.17 1.177643e+04

Even though we’re probably not going to use the total population and total household variables for any analysis here, I kept them because those sorts of variables that define what your observational unit is are important for checking what’s going on in your data. By which I mean a census tract is made up of a bunch of people (usually about 4,000) in a contiguous area who mostly live in households. But if you work with census data enough, you’ll know that some places have population but few households, or only very small populations altogether—a tract might actually be a jail or a set of college dorms, or maybe the majority of a tract is those sorts of group quarters, and the remainder is too small to reliably calculate some of the data. What we want to do with those tracts can depend on context, but I’ll drop them here.

acs_income_tracts <- dplyr::filter(acs_all_tracts, !is.na(median_hh_income))

Typical values

Does anything seem weird about the median household income values? Look back at Figure 1 where it may be more apparent. (We’ll talk about this anomaly in the data.)

Switching to cost burden rates:

ggplot(acs_income_tracts, aes(x = total_cost_burden)) +
    geom_histogram(binwidth = 0.03)

This approaches a normal curve, but is skewed. From the histogram, the mean looks to be around 0.3 (looking back at the summary, this is correct), but with quite a few tracts with higher rates. Because this is a proportion, we don’t expect there to be any values below 0 or above 1.

A boxplot can make it a little easier to figure out what’s typical in your distribution.

Unusual values

ggplot(acs_income_tracts, aes(x = pop_density)) +
    geom_histogram(binwidth = 1000)

There are a few tracts that are extremely dense. If we wanted to get a sense of more typical tracts, we could filter those, either from the data or within the limits of the chart:

acs_income_tracts |>
    dplyr::filter(pop_density < 30000) |>
    ggplot(aes(x = pop_density)) +
    geom_histogram(binwidth = 1000)

acs_income_tracts |>
    ggplot(aes(x = pop_density)) +
    geom_histogram(binwidth = 1000) +
    scale_x_continuous(limits = c(NA, 30000))

We could decide to investigate those high-density tracts. For example, if we’re interested in housing costs, we might drop tracts that seem to mostly be dorms. However, at least these tracts in Montgomery County are actually high-rise condos bordering DC, so we should keep them in.

Covariation

Especially when we talk about housing and socio-economic data, we expect things to be correlated—probably even more so than with naturally occurring phenomena, since so much of where we live and what resources we have are determined by history and policy decisions. So it shouldn’t surprise you to find correlations in data like this. In fact, the CDC PLACES dataset (justviz::cdc) uses demographic data to model health measures where they don’t have direct measurements available, so in cases like that you actually want to lean away from digging into correlations too much, or you might end up just confirming the makeup of the model, not finding anything new.

A categorical and a numerical variable

I’ll reshape the data to get housing tenure into one categorical variable so we can study housing cost burden by tenure. (If the code doesn’t make sense, refer back to the data wrangling notes.)

cost_burden <- acs_income_tracts |>
    tidyr::pivot_longer(
        cols = dplyr::matches("cost_burden"),
        names_to = "tenure",
        values_to = "share"
    ) |>
    dplyr::select(county, name, total_hh, tenure, share) |>
    # extract just the text in each tenure label before the underscore
    dplyr::mutate(
        tenure = stringr::str_extract(tenure, "^[a-z]+") |> forcats::as_factor()
    ) |>
    dplyr::filter(!is.na(share))

head(cost_burden)
county name total_hh tenure share
Allegany County 24001000100 1403 total 0.19
Allegany County 24001000100 1403 renter 0.20
Allegany County 24001000100 1403 owner 0.19
Allegany County 24001000200 1343 total 0.19
Allegany County 24001000200 1343 renter 0.29
Allegany County 24001000200 1343 owner 0.18 
ggplot(cost_burden, aes(x = share, color = tenure)) +
    geom_freqpoly(binwidth = 0.02)

The bit about calling after_stat in the book chapter doesn’t apply here, since we have the same number of observations for each tenure.

ggplot(cost_burden, aes(x = tenure, y = share)) +
    geom_boxplot()

ggplot(
    cost_burden,
    aes(x = forcats::fct_reorder(tenure, share), y = share)
) +
    geom_boxplot()

Two categorical variables

This is a pretty contrived example to match section 10.5.2, but I’ll bin homeownership and housing cost burden into categorical variables, and look at these by count.

acs_income_tracts |>
    dplyr::mutate(
        ownership_brk = cut(
            homeownership,
            breaks = c(0, 0.25, 0.5, 0.75, 1),
            include.lowest = TRUE,
            right = FALSE
        )
    ) |>
    dplyr::mutate(
        income_brk = cut(
            median_hh_income,
            breaks = c(0, 5e4, 1e5, 1.5e5, Inf),
            include.lowest = TRUE,
            right = FALSE
        )
    ) |>
    dplyr::count(income_brk, ownership_brk) |>
    ggplot(aes(x = income_brk, y = ownership_brk)) +
    geom_point(aes(size = n)) +
    scale_size_area()

Two numerical variables

Same variables, without the binning

ggplot(acs_income_tracts, aes(x = median_hh_income, y = homeownership)) +
    geom_point()

The book goes through a few methods to deal with overplotting here, including change alpha (opacity) or using a 2D density.

ggplot(acs_income_tracts, aes(x = median_hh_income, y = homeownership)) +
    geom_point(alpha = 0.2)

ggplot(acs_income_tracts, aes(x = median_hh_income, y = homeownership)) +
    geom_bin2d()

Back to top