Using Tree Maps to Visualize Two Data Dimensions Simultaneously

Tree Maps (sometimes confusingly also known as Heat Maps) is a visualization technique in which data is represented as a series of rectangles whose dimensions are represented by color and size. It originated in displays of the values in a data matrix in which high values were represented by darker colored squares and low values by lighter squares.

Tree Maps are a useful visualization technique when exploring situations in which two variables interact or are interdependent in some way. For example, analyzing profitability by company size by state or the size and number of documents by custodian or document type.

In a logistics environment, when analyzing and ranking the opportunity presented by delivery to a set of zip codes, two dimensions of interest are: delivery area and delivery volume. The larger the delivery area, the longer the travel time, the greater the cost. The larger the volume the greater the profitability because within the limits of carrying capacity, the margin costs of additional deliveries are minimal. Zips can vary widely in land area and so the same delivery volume can represent a good business opportunity in one zip and not in another. Similarly a low delivery volume may be acceptable if the delivery area is very small (e.g. a building or a single block).

The tree map (or heat map) below shows the relationship between the area of a zip and delivery volume.

The land area of the zip is represented by the size of the individual squares: the larger the square, the larger the land area. The color of the squares represents the delivery volume: the darker the color, the greater the delivery volume. From a logistics business perspective; small very dark squares good, large light squares – bad. Using this visualization technique, it is very easy for sales and operational staff to identify which zips are likely to represent the better business proposition.

Several software packages are available which support such visualizations (see http://en.wikipedia.org/wiki/List_of_Treemapping_Software ). The one shown above was generated using LabEscape's Heat Map software (http://www.labescape.com/ ).