Effectiveness criteria for data visualization

According to research performed by Cleveland and McGill there is a difference in the accuracy in quantitative, ordinal and nominal tasks (Mackinlay, 1986). Higher tasks are accomplished more accurately than lower tasks. Tasks to perceive different positions rank as best and differences in shapes are the perceived as being the worst.

How to maximize the effect of an visualization

The following table (adopted from Cleveland and McGill) ranks which task is most effective for each data type (quantitative, ordinal and nominal).



Table: ranking of encoders by effectiveness

The above table does not provide us a way to deal with compound relations within one graphical presentation. A compound relation consisting of two different relations can be composed by using length for the first relation (for instance Sales) and color for the second relation (for instance Product line).

Clarification: the best encoder is position

The following example may clarify the issue: Imagine a real estate agent who wants to see three relations at once: House(Selling price), House(Location) and House( Lot Size ). The attribute Selling price has a quantitative set ordering and according to the Cleveland’s ranking, the best encoder is position. However, the user wants to see the house location plotted on a map. And this can only be accomplished by encoding the location relation by using position. I conclude therefore that the system should know which relation of a compound relation is more important then another. The already used encoders will restrict the use of other encoders of the remaining relations within the same presentation. Further, some quite similar encoders (for instance position and length) should be avoided within one composite presentation. Nowadays, most managers are still men. Nearly 10% of men have problems distinguishing different colors. Therefore, colors should be carefully applied within graphical presentations especially expressing ordinal and nominal information since color ranks high in those type of information.

User Goals and Business Intelligence functions

OLAP technology is an important enabler of Business Intelligence-functions like drill-down, rotating, nesting, slice and dice and so on. In order to satisfy user goals, the information system provides the user with functions that can be executed. An Business Intelligence tool is a special type of information system that is primary concerned with data visualization and data manipulation operations. The mapping between the user goals and the functions of the system can be relatively easily performed in the case of user-task-oriented operational information systems. BI tools are quite different from such information systems. They support the user goals on a very high level where user-task-oriented operational information systems will basically satisfy lower level user goals like tasks and sub tasks. They can be modeled by employing use cases.

Support senior managers with BI tools

The purpose of Business Intelligence tools is to support senior managers in their decision making process, in particular the intelligence phase. The intelligence phase highlights every actual or foreseen deviation of the desired situation (Keuning and Eppink, 1993). This high level goal cannot be instantly and exclusively linked to functions provided by the Business Intelligence system since the state of the data can call for further analysis or not. If no deviation from the desired situation is perceived, the manager stays in the scanning mode until he perceives another deviation or quits the system. Moreover, the senior manager meets a particular situation or problem a few times in his career.

Problems seem to be ill-structured at strategic level

Problems at the strategic level seem to be ill-structured and complex. Employing use cases in order to link user goals with functions and to derive the order of steps that have to be undertaken within the task is difficult to achieve. Furthermore, it is difficult to distinguish whether such functions reside in the application or in the presentation layer of the system. Is the phenomenon of rotation a function that is provided by the application or a feature of the interface? Whatever it is, we inspect such features because they support the high level goals of the managers and it can be seen as a concrete part of the managers task model.