As you work with the Tradeoff Analytics service, you may have questions about specific aspects of the service, its use, and its results. Some common questions and their answers follow. The questions are grouped by topic, and many include links to the documentation for more information.
What is the Tradeoff Analytics service?
The Tradeoff Analytics service helps decision makers to make better choices while taking into account multiple, often conflicting, goals. The service can support users in making complex decisions, like what mortgage to take or which medical treatment to pursue, or common decisions, like which laptop to purchase.
What languages does the Tradeoff Analytics client library support?
The widget is localized to support 13 languages; see the description of the
locale field in The TradeoffAnalytics constructor for a list of supported languages. By default, the library uses the language of the user's browser.
Where can I ask questions and get answers to problems I have with the Tradeoff Analytics service?
Is there a way to rank the objectives for a decision problem?
No, all objectives are equally important. However, you can use filters to set thresholds on objectives' values, modifying the objectives to suit your preferences. For more information, see Filtering objectives.
Does the service support objectives with non-numeric values?
Yes. The service supports four types of objectives:
text. For information about specifying types of objectives, see Specifying objectives; for information about working with objectives in a visualization, see Using the visualization interface.
Is there a limit on the size of the input data set?
generate_visualization query parameter of the
POST /v1/dilemmas method is set to
true, the service imposes the limits described in Generating a visualization. If the parameter is
false, no limits apply.
Does the service accept input in a format other than JSON (for example, online data)?
Currently, the service supports input represented only in JSON format, but the input can be generated dynamically. For instance, assume an application in the retail domain has an initial set of objectives. The application can access a catalog of products and generate one option per product in the catalog. The application is responsible for defining the objectives correctly and for any transformation applied to the objective values (for example, converting units of measure).
The service offers a CSV importer tool that lets you import data from your own Microsoft® Excel file by converting it to JSON. The tool guides you through the process of uploading the data into the service.
How does the service categorize options as best candidates, excluded candidates, or incomplete?
Best candidates have no competing options (also referred to as solutions) better than they are for all objectives; excluded candidates are those for which at least one option is clearly superior for all objectives; and incomplete options fail to specify values for all objectives. For more information, see Understanding candidates.
Can I further refine the set of best candidates to identify those most likely to satisfy my needs?
find_preferable_options query parameter of the
POST /v1/dilemmas method to
true to produce a subset of preferred solutions. The service analyzes the best candidates and chooses the subset that it believes will satisfy the most users, and it provides a score that indicates its confidence in its selections. For more information, see Requesting preferable options.
How can I learn why an option was excluded?
The service lists all excluded candidates by name in the Excluded tab under the Table view in the second step of the interface process. For more information, see Excluded candidates.
How can I learn what data is missing for an incomplete option?
The service lists all incomplete candidates by name in the Excluded tab under the Table view in the second step of the interface process. For more information, see Incomplete candidates.
How do I interpret the
shadow_me fields of a problem resolution?
The service uses these fields to indicate when its analysis determines that one option is better than (shadows) another option. When such a relationship is found to exist between two options, the fields are included in the
Solution JSON that is returned for the options as part of the problem resolution. For more information about interpreting these and related fields of a problem resolution, see Need explanation of Solution response from Tradeoff Analytics API on the Watson forum.
How do I interpret the map visualization?
The map visualization provides a graphical overview of the best candidates for the problem; it omits excluded and incomplete candidates. For more information, see Working with the map visualization.
How do I interpret the line-based, non-map visualization?
The line-based visualization, officially referred to as parallel coordinates, is another type of visual display for the best candidates of the problem resolution. For more information, see Working with the parallel-coordinates visualization.
Are there recommended guidelines to follow during the decision-making process?
The recommended approach is implemented via the Tradeoff Analytics interface, which features a three-step decision-making process. The process follows Ben Shneiderman's rubric, "Overview first, zoom and filter, then details-on-demand" (Shneiderman, 1996.) For complete information about the process, see Using the visualization interface.
Are there best practices for working with the map visualization?
To make a more informed decision and to minimize regret, consider the following guidelines for using the visualization:
Is it possible to format objective values with the Tradeoff Analytics client library (for example, to add a euro symbol to a price value)?
Yes. For a
numeric objective, you can specify the number of decimal places, a currency symbol, a prefix, or a suffix; for a
datetime objective, you can specify the format of the date and time. For more information, see Formatting numbers and dates.