Next: Predictive Analysis Up: COMPARATIVE EVALUATION Previous: Metrics

## Task Scenarios

We use three representative task types of tabular data exploration in our evaluation. These tasks require both a local and a global search of the table [12] to assess special properties of the data presented and build relations among different row and column entries. We define a task taxonomy consisting of three categories.

• Type 1 involves a simple table lookup and the ability to locate a single value in a relatively large data set of information. This corresponds to extracting a simple fact from the data and answering elementary questions [9]. In this case, it consists of associating a data point entry with a row and column.
• Type 2 represents single column or row scanning tasks. This type requires a basic understanding of the data and the ability to find outliers (e.g. minimum, maximum values) and value ranges over a single column or row.
• Type 3 focuses on the ability to perceive a trend in the table and to conduct a global search. It may require the user to look for changes across columns and rows, and to form a mental image of the data. In addition, the user may be required to understand how changes in one variable affect other variables.

Table 1 gives examples for the task taxonomy described.

Table 1: List of Task Scenarios
 Task No. Task Definition Task Type 1 What was the urban population count for New York in 1920? 1 2 What is the region that has the maximum rural population count in 1990? 2 3 Identify the regions that have a urban population count between 10,000,000 and 20,000,000 in 1990 2 4 What is the trend (increase/decrease) in the rural population count for the Northeast Region? 3 5 What is the difference between the Northeast and Midwest regions in terms of urban population count between 1950 and 1990? 3

Next: Predictive Analysis Up: COMPARATIVE EVALUATION Previous: Metrics
Nada Golmie, Bill Kules
1999-12-06

Web Accessibility