Deciding which classification scheme to use

When mapping quantities, you may ask yourself:

· Which classification method should I choose?

· How many classes should I have?

There are no "correct" answers. The best classification scheme for a given map layer depends on the purpose of the map, the characteristics of the data, and cartographic considerations such as how easily the resulting map can be understood.

One approach is to let the data inform your decision. When you are looking for patterns in your data, you could try different classification methods and visually analyze the resulting maps, then select the method that seems best. To evaluate classification schemes before you map them, you can use a graph that ArcMap provides called a classification histogram.

The classification histogram charts the number of attributes (features) for each attribute value. The bottom axis shows the attribute values, and the side axis shows the frequency of the values. The height of the gray bar indicates the number of times a given value occurs in the table (its frequency) .

When deciding on the number of classes, there is one rule of thumb you can use: fewer is generally better. Three to seven classes is usually best.

classification histogram helps you visualize how attribute values are distributed across the overall range of values. The blue lines show the current class breaks (the highest attribute value in each class). The data in this histogram is grouped into three classes.

Another approach is to choose a classification scheme first, and let the attribute values fall where they may. There may be a scientific or statistical reason for using a particular classification method with particular data. Or, you might have predetermined standards or criteria that dictate the method or number of classes.

The table below provides some general guidelines for choosing an appropriate classification scheme.

Classification method	When to use	How many classes to have
Natural breaks	When attributes are distributed unevenly across the overall range of values	Choose a number that best reflects the natural groups of attributes you want to show.
Equal interval	When you want all classes to have the same range	Choose a number that produces an easily understood interval, such as 2, 50, 1000, etc. Or, choose the number of classes that produces a map with your intended message.
Quantile	When attributes are distributed in a linear fashion (an even distribution across the range of values and little variation in the number of attributes for each value)	Choose a number that makes sense for the purpose of your map.
Manual	When you want classes to break at specific values	Choose the number that makes sense with your classification scheme. For example, you would need two classes to show values above and below a certain threshold value.