How to Make a Choropleth Map (Free, No GIS Software)

What a choropleth map is, and when to use one

A choropleth map colours predefined regions by a data value: darker or more saturated areas carry higher values. It is the right choice when your data is already aggregated to areal units (per county, per country, per ward) and the variable is a rate, ratio, or density rather than a raw count. Election results by state, poverty rate by district, rainfall by region, and yield per hectare by county are all natural choropleths.

What you need: boundaries plus a value

Every choropleth is two pieces joined together. The first is the geometry: the shapes of your regions, as a shapefile or a GeoJSON file. The second is the attribute: the number you want to map for each region. Sometimes both live in the same file (a shapefile whose attribute table already has your column). Often the shapes come from one place and the numbers come from a spreadsheet, and you join them on a shared key such as a region name or code.

• Boundaries are free and global. Natural Earth has countries and states with data already attached. geoBoundaries and GADM go down to districts and wards. Your national statistics office usually publishes official admin boundaries.
• Values come from a census, a survey, an administrative record, or your own fieldwork, usually as a CSV or spreadsheet with one row per region.

Step by step

Classification: the choice that makes or breaks the map

Classification decides how continuous values are grouped into colour classes. The same data can look completely different under different schemes, so pick deliberately rather than accepting a default.

• Quantile puts an equal number of regions in each class. Every colour is used and the map looks balanced, but class widths vary, so two regions in different classes can be almost identical.
• Equal interval splits the value range into equal-width bands. It is honest about magnitude and great for evenly spread data, but with skewed data most regions pile into one or two classes.
• Natural breaks (Jenks) finds groupings that minimise variation within each class and maximise it between classes. It usually reads best on real, lumpy data such as population or income.
• Standard deviation shades by distance from the mean, which is ideal when the story is who is above or below average rather than the raw level.

A good habit is to try quantile and Jenks side by side. If your conclusion flips between them, your classes, not your data, are doing the talking, and you should say so. For a deeper treatment, see the guide on choropleth classification methods.

Colour, without misleading

• Use a sequential ramp (light to dark in one hue) for data that runs low to high.
• Use a diverging ramp (two hues meeting at a neutral middle) only when there is a meaningful midpoint, such as a change, an anomaly, or a value above and below average.
• Keep the number of classes to about five. More classes look precise but exceed what the eye can distinguish.
• Check that your ramp still reads for colour-blind viewers. Sequential ramps in a single hue are the safest.

Common mistakes to avoid

• Mapping counts instead of rates, so the map just shows area or population size.
• Letting an outlier stretch the whole colour scale so every other region looks the same.
• Comparing two maps made with different classifications or class breaks.
• Reporting area or distance from a map in a projection that distorts them. For measurement, use a tool that computes real geodesic area and distance rather than raw coordinate units.

None of this requires a GIS licence or an install. Everything above runs in a browser tab. When you are ready, open Groundtruth and drop in a file.