 |
Missings |
|
| Are | ||
| Now | ||
| Equally | ||
| Treated |
|
Missings |
|
| Are | ||
| Now | ||
| Equally | ||
| Treated |
A brief summary of MANET features
First Interactive Implementations of
Boxplots are applicable in nearly all statistical packages. Only few of them allow selection and highlighting inside graphs, non of these offer highlighting in boxplots. MANET offers highlighting inside boxplots by plotting an extra boxplot of the highlighted data upon the boxplot of all data. To distinguish between the two boxplots, the underlying boxplot for all data is drawn in modified way. The whiskers have the same width as the box itself, thus the box is extended to the ends of the whiskers filled with a lighter grey. The figure below shows an example of how easy the comparison of the distribution of a subpopulation is using highlighted boxplots.
Weighting in plots is basically a very easy thing. Every case of the displayed variable is weighted, i.e. multiplied, with its value of the weighting variable.
The above plots of data of 5 US states in the Midwest, show how weighted plots can be used to visualise the quantitative structure of a dataset. The scatterplot in the left shows, that there is basically one county with both high white and black population. Weighting the histogram of the population figures of the whites with those of the blacks, shows clearly, that this qualitative effect is nearly totally concentrated on this one county (which includes Chicago).
But weighting can also be useful of reasons of efficiency. Count data with very many replications of the same combinations of levels of the variables, can be stored much easier when a frequency variable is introduced. This variable can be used by MANET to weight barcharts, histograms and mosaic plots, to ensure the right quantities.
Mosaic plots have already been described above. MANET is the first application which includes mosaic plots in an interactive context. Static representations have been developed for SAS as well as for S-Plus. But the most impressive advantages of mosaic plots in comparison to barcharts arise with interactivity. Highlighting a single box in a mosaic plot corresponds to highlighting the intersections of different categories in various barcharts - which is far more complicated. Thus the user can make a selection of subgroups by single mouseclicks, which would be very difficult with barcharts.
Mosaic plots which display many variables vary much with the change of the order of the variables. The order of the variables of a plot can be easily changed in two ways:
The mosaic plot updates after every keystroke, enabling the user to get a succession of alternative views very quickly. Although the number of bins in a Mosaic Plot is limited to 256, the number of variables is not limited. If the product of the number of categories is more than 256, MANET shows the first k variables with a product of less or equal to 256.
The following figures show the example of the Titanic passengers, survivors are highlighted.
| Sensible order of variables: Class, Age, Sex | Alternative oder: Age, Sex, Class |
 |
 |
Classical methods of graphical data visualization start from low dimensional projections of the data and try to explore and reveal higher dimensional structures by various linking techniques.
This raises the question concerning 'interesting' projections in data sets.
There are several methods from from multivariate statistics for this purpose such as principal component analysis or correspondence analysis. By reducing dimension according to different criteria of optimization these techniques allow to concentrate on only a few factors. But to locate structural anomalies their applicability is limited - as in different data sets dependent on the context different structures are 'special'.
Graphical methods are however extremely useful for drawing such conclusions.
The basic idea of interactive biplots is to combine those approaches.
First of all graphics are a tool to visualize the theoretical methods. They give an impression of how well the method works for the underlying data as well as it reveals the said structural particularities.
The implementation of Biplots into the software MANET shows, that with additional interactive methods such as querying, linking and highlighting a useful working process with this kind of graphic becomes possible.
|
Trellis Displays were introduced by Bill Cleveland in 1993, and are only available inside S-Plus as yet. Although Trellis Displays are controversial, they offer the possibility of achieving a systematic view of the data. Interactivity can be accomplished inside an interactive environment such as MANET offers. Trellis Displays inside MANET are restricted to two categorical conditioning variables, forming the rows and columns of the trellis. Creating shingle variables has been left out because shingling seems to be a dangerous and misleading technique, although the shingling could be done in an interactive manner via sliders. The right example shows a simple trellis display for head injuries split up by year. The cases for airbag-cars are highlighted. The barchart below shows the proportions of the highlighted cases over the years. Single boxplots allow an adjustment of the box-width, according to the number of cases, which is not applicable in trellis displays yet. |
 |
 |
Traces monitor statistics of selected data of different variables, while continously selecting data in a plot. The most common application of traces is to move the brush over a map. The left example shows a trace over the south-east border of Bavaria. While brushing over the different counties, every x pixels or t seconds the statistics are recalculated and are plotted in the time-series like graphs at the right. The parameters x or t can be set by the user. |
