Problem Summary
Faceted analytic applications aggregate, organize, and summarize data from a variety of quantitative and qualitative information sources (such as transactional records, inventories, sales data, human capital information, part and product information, and so on). They allow users to interact with complex data sets to generate insights and “intelligence” to address a broad range of discovery goals via unpredictable, cascading questions. These discovery goals can vary in breadth and complexity, from simple, focused “analytic fact finding” tasks and questions such as:
- “Did we achieve our sales goals for product X last quarter?”
- “Which regions had the best customer satisfaction during the past year?”
to broad, complex, pattern analysis and evaluation tasks such as:
- “I need to understand part obsolescence and manage risks for our products”
The scope and type of interaction can span a wide spectrum of discovery modes, from “Lookup” activities (such as known-item search, fact retrieval, etc.) to “Learn” and “Investigate” activities (such as comparison, aggregation, analysis, synthesis, evaluation, etc.). They allow users to readily and intuitively address the unpredictable “cascading questions” that emerge as they interact with complex information sets to support progressive, serendipitous discovery.
The information summaries and presentations can take the form of data visualizations (charts, graphs, etc.), tabular data presentations or other views selected to optimize the effective communication of key metrics, patterns, trends and overall status. This pattern builds on the design principles established by others in the visualization / information design community (e.g. Edward Tufte, Stephen Few) and focuses specifically on the challenges presented by enabling users to dynamically interact with information visualizations and summaries through facets.
Usages
|
Constraints and Challenges
|
-
Provide a clearly identifiable “Home” or “launch” page that supports the initiation of common discovery tasks and acts as a hub for ongoing discovery work. The optimal design of the “Home” or “launch” page will depend on its purpose and context, but the following are common types:
-
Enable iterative exploration through the display of user interface elements such as:
-
By default, provide information summaries and visualizations of the high priority metrics
-
Provide links to past discovery work (e.g., “bookmarked” views, results sets, saved searches etc.) to enable resumption or recall of past analyses
-
Allow the user to pivot between different views of the data or different levels of detail. For example, switching between components parts vs. assemblies vs. suppliers vs. usages; or between orders vs. products, vs. customers; and so on.). The ability to pivot between different views is particularly beneficial in analytical applications where exploratory modes of interaction are more prevalent.
-
Present the inquiry results at an appropriate level of abstraction. In some cases it may be appropriate to present results at their most granular level, i.e. as individual records. However, in other cases it is more appropriate to aggregate records by “rolling them up” into meaningful groups or higher level entities. Determine the appropriate level of aggregation for a particular view by considering the overall search and discovery context, i.e. the user, their goals/scenarios, the information assets and the anticipated modes of interaction.
-
This is particularly important in analytic applications due to the multiple ways in which faceted refinements may be applied (e.g. by drilling down on interactive elements, etc.) The effects of these actions – which may be functionally equivalent to applying facet values -- and the interdependencies between them may not be immediately apparent, in which case clear and consistent messaging regarding the consequences of the user’s actions and the current inquiry state is vital. This messaging should provide the dual purpose of providing:
-
Clearly convey the scope of interface controls and their effect (e.g. does applying a given refinement apply across all elements within a view, a single visualization, or a group of elements) through visual affordances (grouping, proximity), text, etc.
-
The guiding principle for the selection and design of information visualizations should be the efficient communication of information to support user insight and serendipitous, iterative exploration. This requires an understanding of the overall search and discovery context, and typically involves consideration of the user, their goals/scenarios, the information assets and the anticipated modes of discovery. Visualizations may include various controls that allow the user to view the data in the manner that best supports their needs. For example:
-
Allow users to refine using via interactive elements within the visualizations. For example, a user currently viewing a set of data grouped by country may choose to “drill down” on a particular country by selecting the relevant element in the visualization. The effect of this action is the same as selecting the equivalent value in a faceted navigation menu: the navigational context is refined to the selected country.
-
Allow the user to pivot between different views of the data for specific use cases. For instance, in a dataset that represents sales of a range of products over a given time period, users may want to pivot between an organization-centric view, a product-centric view, or a customer-centric view, etc. depending on their goals and scenarios.
-
Allow the user to independently control the visibility of selected elements of the display, i.e. the visualizations, results table, faceted navigation, etc. For example, a user may wish to hide the results table to allow further space for exploring high-level patterns in the visualizations, or conversely may wish to hide the visualizations in order to allow further space to view individual records, etc.
Comments (0)