The following guide describes how to configure Structure and Jira to get the most out of Structured Predictions.
You must have Structure 5.5 or higher installed.
Configuring Structure Data Sources
Structured Predictions uses Structure for Jira from ALM Works to define data sources used for predictions. Configuration options allow for some flexibility in how the Structure data sources are configured, but some guidelines will make it easier to align your data with the Structured Predictions configuration options.
There are two primary data sets Structured Predictions needs to be able to identify:
- The issues that serve as estimates (typically Epics) - these represent the more abstract requirements that will group the scope level into "kinds of work" a team will need to do in order to satisfy the requirements for a delivery.
- The issues that identify scope (typically Stories) - these represents the specific units of work an engineer will deliver in a sprint, or as part of the normal process flow if you are using Kanban or some other practice.
A structure that is configured to show Epics as estimates and Stories under those epics to show refined scope is in good shape to feed directly into Structured Predictions. A structure configured this way will look something like the figure below.
Configuring Structured Predictions
When you create a new Structured Predictions gadget, you will specify a number of configuration elements, including how to identify and interpret the information discussed above.
1. Scope Structure
Select the structure that contains the work you want to track.
2. Training Structure
Select a structure that contains one or more completed deliveries. This structure will be used to train the Structured Predictions algorithms as to how the teams working in the Scope Structure have performed in the past. This structure should contain work representation for all teams working in the Scope Structure. It can be the same structure as the scope structure, but if the scope structure does not have any data or only has a little data the results will be highly variable.
3. Training Period
Select whether to use all historical data or just the last few weeks worth of data to train the algorithms. If only the past few weeks are desired, make sure to set the Training Structure equal to the Scope Structure. This will show how the teams have been performing in recent history and how that performance translates into target dates for the delivery.
4. Estimation Level
Select what issue types your organization uses to specify a work definition that is larger than an Agile Story. This is usually an Epic, but if your organization uses other issue types, make sure to include them here.
5. Scope Level
Select what issue types your organization uses to indicate discrete work a developer can deliver. On Agile teams, this is usually a Story and a Bug issue type. If your teams use other issue types, indicate them here.
6. Team Attributes
Indicate how Structured Predictions should identify team backlogs in the data sets defined by the Scope and Training Structures. For example, if each of your teams has its own project, than indicate "Project". If you use a custom field, indicate which one. If you use the built-in Jira component field, select that. If you have a more complex method, you can use a Structure Formula to isolate the team backlogs in the Scope and Training Structure data sets by selecting "Formula" and specifying the proper formula code snippet.
7. Target Date
Specify what date the delivery is expected to be delivered.