|
Flexible Hierarchy
.
Why flexible hierarchies?
Some data is best presented with a fixed hierarchy. For example, census data falls nicely in a hierarchy of country, states, counties,
and blocks which could not be rearranged. Some other data could be rearranged using various hierarchies based on the attributes available.
For example, if you have health data provided for the 2 sexes, 3 races and 5 age groups, you could present the data in a variety of ways.
One example would be by separating men and women at the 1st level, then splitting by race at the second level and then by age. Another hierarchy could be to split by age first, then sex, then race. Flexible hierarchies let you specify the order you want those attributes to be used when creating the hierarchy.
Hierarchy tab: to create hierarchies
- If the data file you are using has a default hierarchy (see creating your own data file), you can 1st remove the default hierarchy. All nodes will appear together on the treemap without any hierarchy. Now you can create you own hierarchy.
Choose from the list of available attributes (at the bottom), then click on the "Add" button and the attribute will be added in the Hierarchy List.
- Attributes that have only a few values (e.g. sex or race) make good hierarchy attributes. Other like "age" need to have their values grouped in bins to split the tree nicely, e.g. you can split between young and old people by binning the age attribute into 2 bins. TO BIN an ATTRIBUTE: click on the right most cell of the corresponding hierarchy attribute table.
For details and tips on using the binning widget please refer to the section on User Defined Bins in Color Binning for Numerical Attributes.
To remove an attribute, select the attribute and click on the "Remove" button.
- You will find that it is better to use put the last/bottom attribute as the label attribute and remove it from the hierarchy definition.
You can create different permutations of the attributes in the Hierarchy List by using the "Up" and "Down", or "auto" to permute automatically and review the many options. Treemap redraws the tree as you change the hierarchy.
- Once you are happy with a hierarchy you can give it a name (by typing a new name where it says "hierarchy 1") and you can use the "New" button to create another hierarchy. Remember to save the settings as well!
|
Example of Flexible Hierarchy
- Open the "election.tm3" file and color by the attribute "Party in 2000". This shows which states voted for who in the last presential election. Changing one of the two colors will make things more readable.
This file has a default hierarchy of states within regions, but now you can go the newly added "Hierarchy" tab window and add levels in the hierarchy.
For example, you can select the attribute "Party in 1996" from the list of attributes and then click on "Add".
The hierarchy now has an additional level, and regroups all the states that voted for Democrats and those who voted for Republicans.
If you now select "Party in 1996" in the hierarchy and click on "Move Up", this particular attribute will be at the top level of the hierarchy.
Next, remove the attribute "Party in 1996" from the hierarchy (by selecting it in the hierarchy and clicking on "Remove").
Add the attribute "Population" in the hierarchy. Since it is a numerical attribute, there is no natural grouping. The bottom binning widget shows the distribution of the values.
Take a look at the hierarchy panel for the attribute "Population". It says 49 (for 49 values) and "No binning".
No binning is actually a drop down box with different options, namely, Equally Spaced Bins, Equally Dense Bins, and User Defined Bins.
|
