Further Analysis of TT-Plot Data

In the previous sections I often used a visual language using colors to make communication easier. It must not be forgotten that the color schemes used for presenting the TT-plots are only a representational feature. They always represent measurements extracted from the original data in the form of distances or angles which finally build up the matrices called TT-plots.

The aim of the Time-Plot family is to provide fast methods for recognizing and pinpointing special features in large datasets. After a first inspection the analysis can be refined and extended in several ways:

• zooming in by using a smaller time frame
• reselection of values within a TT-plot (defining values of interest)
• reclassing the values into categories
• performing transformations on the values (e.g. logarithm)
• building histograms of values
• calculating statistical values from the matrix (mean, variance etc.)
• Fourier analysis of the TT-matrix

This list is not complete but indicates in what directions further analysis of these plots may lead. Examples of this kind are given in chapter 6 together with the presentation of applications to biological data.

The above sections introduced two kinds of TT-plots, the TT-$\delta$ plot and the TT-$\pi$ plot which describe distance measurements and parallelity aspects, respectively. These are only two aspects that can be used for a temporal analysis in TT-plots. The creation of other TT-plots can be more or less performed in a similar manner as above. For example the following characteristics could be used for other TT-plots:

• altitudinal difference
• differences in attribute data angle difference from a specific direction (e.g. north (implemented in the prototype), or valley main direction)
• relative speed
• derivatives from the above (speed change, direction change etc.)

The list is not intended to be comprehensive but instead to give some ideas for further developments of this technique.