Heatmap

A heatmap (English: heat = heat ',' heat ', map = map ' ) is a diagram for visualization of data whose values ​​depending on a two-dimensional set of definitions are represented as colors. It is used to detect quickly and intuitively particularly striking values ​​in a large amount of data.

The term heat map is blurred. He reaches into the narrow interpretation of ( high ) temperatures on a map to Tree Maps with unsystematic color coding. Widely used, the provisions

• Two-dimensional definition Quantity: rows and columns of a table, axes of a mathematical function, image ( screen ) coordinates (photo, web page)
• Color coding of the dependent values: representation by a temperature scale ( which is recognizable as such), or stepped ordinalskalig assignment of hue, a (part) of a spectrum or a color circle following; monochrome variation of saturation or brightness; but also black-and- white or gray- values ​​, shaded, dotted, hatched or graphically distinguished in other ways.

The name Heatmap derives from the fact that some colors are associated with temperatures.

• 4.1 Bubble Chart
• 4.2 Column Chart

History

The use of heat maps goes back at least to Toussaint Loua, which represented in 1873 statistical data of the arrondissements of Paris as a table with shaded areas, only black and white, then in color.

Sneath (1957 ) represents the result of a cluster analysis by permuted rows and columns of a matrix to place similar values ​​near each other according to their accumulation. The idea to add cluster trees of rows and columns of a data table, goes back to Robert Ling in 1973. Leland Wilkinson developed in 1994 the first computer program ( SYSTAT ) to produce cluster heat maps with great color depth.

Color coding

If the dependent variable is a scalar, so they can be mapped naturally to a temperature scale. If assigning a second step, a temperature a color to blue is considered a cold color, red, orange and yellow as warm colors, it will declare the name of the heatmap.

The combination of color and temperature is culturally different and not mandatory. The choice of the color scale is arbitrary and therefore exempted the creator of the heatmap.

With fixed end points of the scale, there is the issue of transitional colors. A light bulb with increasing current flow glows only red ( warm), then yellow and finally white ( hot). Supplemented by the cold blue, a color scale look as follows: black ( achromatic, coldest ) - Blue - Purple - Red - Yellow - white ( achromatic, callest ).

A wider use of the color circle progresses from blue through green - yellow - orange to red before. Purple can be added here appear as coldest color below blue. Red then the hottest, the coldest violet color, as opposed to incandescent model where violet and red represent mean temperatures.

Color is a fuzzy term in this context. It can be even more specific about by hue, saturation and brightness, or by red, green and blue components, or other parameter models. This either further dimensions (data series ) can be represented in a diagram or defuse by limiting problems such as color blindness or black and white printing.

Variants

Heat map

A heat map in the strict sense is a temperature distribution on a map, such as the (summer ) temperatures in the city or on a topological subway map.

Heat map is the literal German translation of heatmap. To what extent thus described the same concepts or False Friend is created, whether the borrowed word may lay down in its English or German spelling in the German language, is unclear.

Thermal imaging

A thermal image represents temperatures are color-coded in two dimensions. Since the measured value is in principle already a temperature eliminates the abstraction step of reading to a temperature scale. Thermal images thus represent the original form of the heat map and explain its name.

False color image

In false-color images, a color or gray levels of an original image to new colors recoded to increase the contrast or the intuitive understandability.

Clustered heatmap

If there is no order on the axes of the table, the rows (columns ) can be permuted without loss of information. If the permutation performed so that similar rows (columns ) follow each other, so the table will receive additional structure and the cluster fall clearly into the eye.

Choropleth map

In a choropleth map the areas of color no rigid rectangular grid, but the irregularly shaped areas on a map, follow assigned as vegetation types, climates or administrative boundaries. Some use the more concise Heatmap erroneously instead of the correct term choropleth map for reasons of convenience

Tree Map

A Tree Map is not based on a two-dimensional table, but on a hierarchical set of nodes. It is doubtful to what extent the concept of heatmap is justified here, especially if the colors do not follow intuitively understandable system or trivial values ​​such as the path length is color coded instead as the partition size in a file system. The same is true for mosaic diagrams.

Alternatives

Instead of using color value of a variable can also be represented by other characteristics. Combinations with colors in general, and a heatmap in particular, are possible.

Bubble Chart

Instead of a color of the dependent variable value can be represented in a bubble chart by a geometric shape such as a circle. The value is then optionally proportional to the diameter or area. For a table with three independent parameters, a ball out of the circle. Will use this representation particularly when the axes organized and the table is sparse.

Column Chart

In a column chart, the variable values ​​are represented by the height of columns. Additional dimensions can be obtained by adding more columns per table field. Maxima can be captured very well, smaller values ​​in the background are obscured by larger in the foreground. The same applies to a surface chart.

Applications

A selection of applications for heat maps ( list not exhaustive):

Web Analytics: click density on a web page

Stay on the football field

Combination of heat map and surface plot

Software

Properties

People with color blindness can use the information in a heatmap with bright colors worse or not at all perceive, contrary to the intention of the creator. This should be taken into account when considering the accessibility.

Criticism

The output medium, the color can vary: Each screen provides a color other group; when printing takes place a change from additive to subtractive color mixing, which changes the color impression.

To correct the situation, the restriction to monochromatic color scale or gray values ​​here.

Sometimes heat maps are considered to be pretty, but useless. This criticism, however, is applicable to any visualization that is not used in a targeted manner.