Automatic Rankings

Owned by Nicolás Palombo (Unlicensed)
Last updated: 18 Sept, 2008Version comment
3 min read
Unknown macro: {scrollbar}

Objective

The objective of automatic ranking is to improve the response time that the user experiences, by providing a partial view of the data, in order to remain within certain defined limits.

This is accomplished by executing a set of predefined rankings on the dimensions involved in the analysis. It is possible to specify a filter for each dimension in the axes to reduce the number of elements involved.

When O3 detects that an axis will hold more than the predefined number of elements (specified in rows and columns with the properties o3.client.maxXSize and o3.client.maxYSize), O3 Server automatically applies rankings on the dimensions in the axes.

The model can specify, for each dimension, whether this ranking will be activated or not, as well as the number of elements to display.

All four construction modes (See Construction Modes) allow the specification of this Ranking criteria.

Configuring the Automatic Ranking

O3 Server applies automatic rankings considering dimensions in the following order:

  • If there is a user-defined ranking for any of the dimensions, that ranking is considered first (current version allows users to specify rankings for one dimension only).
  • Then the the rest of the dimensions are considered starting from the dimension closest to the numbers.

The order in which the dimensions are considered is very important since this order tries to respect the number of elements indicated at design time.

Automatic Rankings are activated from the Explore menu in O3Browser or from the Options menu in O3Portal. This option is inactive by default.
This default behaviour can be changed setting the property

o3.force.restrictions.ranking=true

in GServer_custom.properties.

Current version supports automatic ranking in one axis only (columns or rows). This means that, even if both axes contain large sets of data, only one of them will be automatically ranked.

When ranking, it is possible to show a summary of the elements left out in the ranking in a special element called "Others".

Example:

Suppose there are two dimensions: Countries with 219 elements and Products with 98 elements.

  1. When nesting Products inside Countries, the size of the axis rises to 21462, exceding the default limit of 2500. Therefore an automatic ranking is applied showing the first 25 Countries.
    Notice that the ranking is applied to Countries because it is the outter dimension in the axis.
  2. If Countries had some ranking criteria specified in the model, stating for instance to display the first 10 elements, then this ranking would be applied showing the first 10 countries for the 98 Products.
  3. If both Countries and Products had some ranking criteria specified in the model, this ranking would be applied when the limit is reached, reducing the number of elements in the outter dimension.
Unknown macro: {scrollbar}