Choropleth Classification
- Introduced by
- Thematic cartography tradition (class-interval methods synthesized by Slocum et al.; Jenks's optimal method)
- Year
- 1967
- Type
- Procedure for grouping data values into ordered classes for a choropleth map
- Family
- Cartographic data classification
What does it do?
Choropleth classification is the cartographic procedure of grouping the values of a quantitative variable into a small number of ordered classes so that areas can be shaded on a thematic map.
Because a continuous distribution must be reduced to a handful of colour categories, the choice of how many classes to use and where to place the break values strongly shapes the map's message — the same data can look uniform or sharply divided depending on the scheme. Standard methods include equal interval, quantile, Jenks natural breaks, standard deviation, and head/tail breaks, each making different assumptions about what pattern the map should reveal.
The intuition behind the method
A choropleth map can show only a few shades, but the data behind it run across a continuous range, so somewhere you must draw lines that say 'everything in here is one colour'. Where you draw those lines is a decision, not a fact: put the cuts at round numbers and you get equal intervals; put them so each class holds the same number of areas and you get quantiles; put them in the natural gaps of the histogram and you get Jenks breaks. Each choice tells a different story — equal interval exposes skew, quantiles guarantee colour variety, Jenks respects clustering in the data. Choosing well means matching the break method to the distribution and to the point you honestly want the map to make.
History
Choropleth mapping dates to the early nineteenth century, with Charles Dupin's 1826 shaded map of education in France usually cited as the first, but the problem of how to choose class intervals became a formal concern of academic cartography in the twentieth century. George Jenks, working at the University of Kansas, articulated the data-model concept of statistical mapping in 1967 and developed the optimization that bears his name, the natural-breaks method that minimizes within-class variance, later made exact by the Fisher dynamic-programming algorithm. Through the latter half of the century textbooks codified the competing schemes — equal interval, quantile, standard deviation, and natural breaks — and Slocum and colleagues' Thematic Cartography and Geovisualization became a standard synthesis. More recent work, such as Bin Jiang's head/tail breaks, has added methods suited to the heavy-tailed distributions common in geographic data.
When to use it
Use choropleth classification whenever you map a quantitative attribute of enumeration areas — population density, unemployment rate, vote share, disease incidence — and must reduce it to discrete shaded classes. Equal interval suits data spread evenly across their range or where round, comparable breaks matter; quantile suits ranking and ensuring every class is visible; Jenks natural breaks suit data with genuine clusters you want the map to honour. It is the right tool when areas are the unit of observation and a single variable is mapped; it is less appropriate for raw counts (which should be normalized first), for variables better shown with proportional symbols or dasymetric methods, or when comparing several maps that must share an identical, fixed classification.
Limitations and considerations
- The map's appearance and message depend strongly on the scheme and number of classes chosen by the cartographer.
- Any classification discards within-class variation, so areas of quite different value can share a colour.
- Jenks and quantile breaks are data-dependent, making maps of different datasets or dates hard to compare directly.
- Choropleths inherit the modifiable areal unit problem: the result depends on how the enumeration areas are drawn.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Turns a continuous distribution into a readable map with a manageable, perceptually distinct set of shades.
- Offers a family of schemes — equal, quantile, Jenks, standard deviation — matched to different data shapes and goals.
- Jenks natural breaks give an optimal, data-faithful classing that respects clustering and gaps.
- Well supported and largely automated in every GIS and statistical mapping package.
Applications
- Mapping socioeconomic rates such as unemployment, income, or educational attainment by area.
- Showing population or housing density across administrative units.
- Visualizing election results as vote-share choropleths by district.
- Displaying epidemiological measures like disease incidence or mortality rates across regions.
- Communicating census and survey indicators to the public in atlases and dashboards.
How it works
How it works
The first decisions are how many classes to use and which classification scheme to apply. Most maps use four to seven classes, enough to show pattern without overwhelming the eye or exceeding the number of distinguishable shades. The scheme — equal interval, quantile, natural breaks, standard deviation, or head/tail — should suit the data distribution and the map's purpose, because it determines whether the map highlights outliers, even ranking, natural clusters, or departures from the mean.
Formula
k \in \{4,5,6,7\}\ \text{classes};\ \text{scheme} \in \{\text{equal, quantile, Jenks, std-dev, head/tail}\}Equal interval divides the range from minimum to maximum into k classes of identical width, placing breaks at the minimum plus successive fractions of the range; it is intuitive but lumps most areas into a few classes when the data are skewed. The quantile method instead sets breaks so that each class contains the same number of areas, guaranteeing a full spread of colours but potentially placing very similar values in different classes and very different values in the same class. They answer different questions — equal width versus equal count.
Formula
\text{break}_j = \min + j\,\frac{\max-\min}{k}\quad\text{(equal interval)}The Jenks natural-breaks method seeks breaks that minimize the variance within classes while maximizing the variance between them, so that each class is internally homogeneous and the classes are well separated. The Fisher-Jenks algorithm finds the optimal such partition for a given number of classes. This respects clustering and gaps in the actual distribution, generally producing the most faithful classed representation of the data, at the cost of breaks that are not round numbers and that change if the data change.
Formula
\arg\min \sum_k \sum_{i\in c_k}(x_i-\bar{x}_{c_k})^2Before symbolizing, raw counts should be normalized to rates or densities — cases per capita or per unit area — because shading raw counts on a choropleth makes large areas look important simply for being large, an area-size bias. The classed variable is then assigned a colour ramp: a sequential ramp for ordered data running low to high, or a diverging ramp when there is a meaningful midpoint, with hue and lightness chosen for perceptual order and accessibility.
Formula
\text{rate}_i = \frac{\text{count}_i}{\text{area}_i\ \text{or}\ \text{pop}_i}Common pitfalls
- Mapping raw counts instead of rates or densities, so large areas dominate purely because of their size.
- Letting the default scheme in the software stand without checking that it suits the data distribution.
- Using too many classes, exceeding the number of shades a reader can reliably distinguish.
- Comparing two maps with independently computed breaks as if their colours were on the same scale.
Frequently asked questions
How do I choose between equal interval, quantile, and Jenks natural breaks?
Match the scheme to the data and the message. Equal interval is best when values are spread fairly evenly or when round, comparable breaks are wanted, though it can leave most areas in one or two classes if the data are skewed. Quantile guarantees that each class holds the same number of areas, ideal for ranking and ensuring colour variety, but can split near-identical values. Jenks natural breaks find the cuts that best respect clusters and gaps in the distribution, usually giving the most faithful map, at the cost of irregular, data-dependent break values.
Why should I normalize counts before making a choropleth?
Because choropleth shading is read as intensity, and raw counts make large areas appear more intense simply because more of anything tends to occur in a bigger or more populous unit. Mapping a rate or density — cases per capita, people per square kilometre — removes that area-size bias and shows the underlying intensity that the map is meant to convey. Raw counts are better represented with proportional symbols than with shaded areas.
How many classes should a choropleth map have?
Usually between four and seven. Fewer than four classes throws away too much detail, while more than about seven exceeds the number of shades most readers can reliably tell apart, especially in a single hue. The exact number depends on the data's spread and the map's purpose; when several maps must be compared, the same number of classes and the same break values should be used across all of them.
Related methods
Sources
Slocum, T. A., McMaster, R. B., Kessler, F. C., & Howard, H. H. (2009). Thematic Cartography and Geovisualization (3rd ed.). Pearson Prentice Hall, Upper Saddle River, NJ.
ISBN: 9780132298346Copied.