Carpet plot
A carpet plot is any of a few different specific types of diagrams. The more common plot referred to as a carpet plot is one that illustrates the interaction between two or more independent variables and one or more dependent variables. One feature that distinguishes a carpet plot from an equivalent contour plot or 3D surface plot is that a carpet plot facilitates interpolation. Furthermore although it is a two dimensional plot, a three dimensional impression of the data can be achieved. A carpet plot can capture the interaction of up to three independent variables and two dependent variables and still be easily read and interpolated.
Carpet plots have common applications within areas such as material science for showing elastic modulus in laminates,[1] and within aeronautics especially for the use of trade studies.[2][3]
Another plot referred to as a carpet plot is the temporal raster plot.
Variants
Cheater plot
A carpet plot with two independent variables and one dependent variable is often called a cheater plot for the use of a phantom "cheater" axis in lieu of the horizontal axis. As a result of this missing axis, the carpet plot's values have to be shifted horizontally. For a better readability the intersections can be lined up vertically and folded carpet plots can be unfolded. [4]
A carpet plot with a filled contour
To add more information to the carpet plot, a filled contour plot can be added as well. The contour lines are a function of the two independent variables. Contour plots can be added to real carpet plots as well as to cheater plots.[5]
Lattice plot
If a third dependent variable should be visualized it is possible to plot multiple carpets for discrete values. If the intersections are connected, a lattice appears. For an intermediate value a whole carpet can be interpolated using the lattice lines.[2]
Temporal raster plots
Another visualization that is sometimes called a carpet plot is a graphic representation of occurrences in a certain temporal relation. Each occurrence is registered in a Cartesian coordinate system, in which both axes show time, have different time resolutions: one axis shows slices of data, the other some sensible interval. A common example would be that one axis shows hours in a day, the other days in a year.
In a 2D plot, the value to be plotted is coded with an intensity or a color. In the 3D variant of the plot, it is often coded as a height.[6] When visualized, particularly the color-coded variant of the plot may easily show a carpet-like pattern.
Carpet plots make it easy to show time-based relations within a large sets of time-interval data and often make it easy to recognize local maxima and minima. Assuming the chosen time division is related to the events, it is also easy to recognize global and local patterns, such as recurrent events.
In the following example, the data is one year's worth of measurements of the outdoor temperatures in Augsburg, with four samples taken per hour. In the according carpet diagram, each column represents a day in the year and contains the values for that day (from 0:00 until 23:45). Despite the high number of measure points (about 35000) it is easy to disdistinguish local and global patterns.
Note the similarity to beat graphs as described by Werner Van Belle.[7]
References
- ^ "Laminate Carpet Plots". Retrieved 6 January 2010.
- ^ a b "Use of Carpet Plots to represent functions of two variables". Retrieved 2 November 2013.
- ^ Raymer, D. P. (1992). Aircraft Design: A Conceptual Approach. Reston: American Institut of Aeronautics. p. 529.
- ^ "'The Generation of Carpet Plots'" (PDF). Retrieved 1 November 2013.
- ^ "'Gasturb Tutorial 2 - Cycle Design Parametric Study'". Retrieved 2 November 2013.
- ^ "'Whole building electrical power over time'". Retrieved 17 December 2008.
- ^ "BPM Measurement of Digital Audio by Means of Beat Graphs & Ray Shooting". Retrieved 17 December 2008.
External links
