Data Classification For Choropleth Maps

The goal of data classification is to put places with similar rates in the same class. When you perform a classification, you group similar features into classes by assigning the same symbol to each member of the class. While this technique allows you to spot patterns in the data more easily, it can also create very different-looking maps. The maps below are the results of five different classification techniques used on choropleth maps.

Natural breaks is an optimization method for choropleth maps which arranges each grouping so there is less variation in each class or shading. It is a method of manual data classification that seeks to partition data into classes based on natural groups in the data distribution. Natural breaks are data-specific classifications and not useful for comparing multiple maps built from different underlying information. Natural breaks classification will minimize variation in each group, maximize between-class differences, and still manage to group outliers in their own class.

The equal interval classification divides the classes into equal groups (e.g. 4-8, 8-12, 12-16). This means that equal interval choropleth maps almost always result in an unequal count (e.g. 200 in group 1, 5 in group 2).This option is useful to highlight changes in the extremes. It is probably best applied to familiar data ranges such as percentages or temperature.

If you want the count in each group to be close to equal you should use a quantile map. Quantile maps try to arrange groups so they have the same quantity and as a result, the shading will look equally distributed. Classes at the extremes and middle have the same number of values. Because the intervals are generally wider at the extremes, this option is useful to highlight changes in the middle values of the distribution.

You should create classes manually if you are looking for features that meet a specific criterion or if you are comparing features to specific, meaningful values. With this classification you can manually specify the upper and lower limit for each class which can be a useful technique for isolating and highlighting ranges of data. For example, if your dataset has an overall range of 0.0465 to 0.1736 and you want to isolate the higher values, you could manually assign all values below 0.15 to one class and all values above to a second class.

These maps avoid the messy (and nearly always imperfect) problem of having to lump the data into classes. Choropleth map data classification is a very powerful form of data filtering that drowns out important details on the map and is easily abused to change what the map says. Unclassed maps let the data speak for itself, and allow even subtle differences between places to emerge as subtle differences in colour. Each unique data value gets a unique colour. They shouldn't be used if you need to get numbers off the map, or need to carefully compare one location to another.