自分のデータから地図を作る方法

このストーリーはまだ翻訳されていません。英語の原文を表示しています。

Most research data that ends up on a map starts as a list. One row per country, per district or per survey site, a column that says where, and a few columns of numbers. This manual takes such a list from a spreadsheet to a finished, publishable world map in GeoCanvas, with no GIS software and no boundary files to find.

The example is real. In January 2025 the TISP consortium published a survey of 71,922 people in 68 countries on how much they trust scientists, and put the data online. The paper shows the country means as a chart (its Fig. 1); here they become a world map.

World map of mean trust in scientists in 68 countries, classed in five colours from yellow to dark blue; countries not surveyed in grey — Where we end up. Weighted mean trust in scientists on a scale from 1 (very low) to 5 (very high), in five classes with the same number of countries each (Equal count); grey countries were not surveyed. Equal Earth projection. Data: Cologna, Mede et al. (2025), CC BY 4.0. Boundaries: Natural Earth.

The build

1Get the table into shape

GeoCanvas needs one row per place and one column that says which place it is. For countries that column can hold names in English or in many other languages, or ISO codes: three letters (DEU), two letters (DE) or numbers (276). Codes are the safer choice, because a code cannot be spelled two ways. Every other column is a value you can map.

iso3,country,trust
EGY,Egypt,4.30
IND,India,4.26
…
ALB,Albania,3.05

Save it as CSV, or keep it as an Excel workbook: GeoCanvas reads the first sheet of an .xlsx. A decimal comma is fine, and so are the usual markers for a missing value, such as NA or an empty cell. They stay missing; they never become zero.

2Upload it and check what was placed

On the map, open the Edit tab and choose Upload file. For a table, GeoCanvas asks where each row belongs. Leave it on In a country, given by name or ISO code and check the column it picked: it chooses the one that places the most rows.

Read the grey box before going on. It counts the rows that were placed and lists every one that was not, with the value that failed to match. A missing country here is a missing country on the map, so fix it in the table and upload again rather than carrying on.

The Place your table on the map dialog, with a preview of the table, the ISO code column chosen and the count of rows placed — The dialog previews the table and says how many rows found a country, before anything is added.

3Add it to the map

Add to map creates a new layer from the country boundaries GeoCanvas ships with (Natural Earth, 1:50m). Every country is there, not only the ones in your table, so that a country without a value can be shown as no data instead of disappearing.

This screenshot was taken in Mercator on purpose, because the next step is to leave it.

The new layer on a Mercator map: every country in light grey with white borders — Every country arrives, light grey. The ones in the table carry their values; the rest carry none.

4Switch to an equal-area projection

A map that colours countries by a value asks the reader to compare areas of colour. Mercator makes that comparison meaningless: it draws Greenland as large as Africa, which is fourteen times bigger, and inflates every country far from the Equator. Pick Equal Earthin the Projections panel. It keeps every country’s area in proportion and still looks like the world people know. Mollweide is the other common choice.

The same layer in the Equal Earth projection — Equal Earth. Africa is back to fourteen Greenlands.

5Colour the countries by their value

Open the layer (it is named after your file), then Symbology. Under Fill choose Graduated and pick the column to map. Only numeric columns are offered, which is a quick check that GeoCanvas read your numbers as numbers.

Then decide how the values are split into classes. Equal count (quantiles) puts the same number of countries in each class, which shows rank well. Equal range (equal intervals) gives every class the same width on the scale, which suits a scale with fixed, meaningful steps. Natural breaks (Jenks) looks for gaps in the data. There is no neutral choice, so state the one you used in the caption, as the figure on this page does.

The colour ramps on offer are the ColorBrewer palettes, made by the cartographer Cynthia Brewer for thematic maps (Harrower & Brewer, 2003). For a quantity that runs from low to high, use a sequential ramp; keep diverging ramps for data with a meaningful midpoint. We used YlGnBu, whose lightest class still stands out against the grey of the countries without data.

The Symbology panel with a graduated fill on the trust column, five classes, and the countries coloured — A graduated fill: the column, the classification method, the number of classes and a ColorBrewer ramp.

6Add a legend

Go back to the layer and open Legend. Switch on Show legend and give it a title that names the variable. The legend states every class boundary and a No data entry, which is exactly what a reader needs to check the map against the table.

The Legend panel with Show legend on, and the legend on the map listing five classes and No data — The legend lists each class with its limits and how many countries fall in it, and counts the countries without data.

7Export for the paper

Open Export. For a journal figure, SVG keeps every border and letter as a vector, so it stays sharp at any size and can be finished in Illustrator or Inkscape. If the figure is done, PNG with the A4 or A3 300 dpi preset is ready to submit, and TIFF (CMYK) is for a print shop. The legend and the source credits go into the file with the map.

Save the map too, with Save to Filein the same panel. The project keeps the projection, the classes, the colours and the data, and the layer’s Layer info records which file it came from, when, and which rows were not placed.

The Export panel with SVG selected — SVG for a figure you will finish in a drawing program; PNG at A4 or A3, 300 dpi, for a figure that is done.

8When the data changes

Data gets corrected, and a revised paper needs a revised map. Upload the new version of the table (same file name or the same columns). GeoCanvas remembers the choices you made last time and offers to replace the values in the layer you already styled: the projection, classes, colours, legend and labels stay as they are. If the new values are wrong, undo puts the old ones back.

The table dialog for a second upload of the same file, offering to replace the values in the existing layer — Uploading the same table again offers to replace the values and keep everything else.

Where this stops

GeoCanvas maps one value per place. A table with a row per country and year needs the year you want picked out before it is uploaded; otherwise the first row for each country is used, and the dialog says so.

Class breaks are computed from the data. Breaks you type in yourself, to match a figure in another paper exactly, are not available yet.

The built-in boundaries are countries, as Natural Earth draws them: de facto control, not legal claims. For districts or regions, upload their boundaries first as GeoJSON or a zipped Shapefile, then the table; the dialog then offers to match rows to that layer by a shared code or name.

And the map draws the numbers it is given. Survey weights, confidence intervals and whether two neighbouring classes differ significantly are questions for the analysis, before the map.

The Maps for research help page summarises what GeoCanvas does to keep a thematic map honest.