Maintenance of Models

Below are a list of best practices regarding maintaining applications, assets, and pages.

• Ensure nodes are “Scoped” correctly. This means, for driver models, that the majority of the nodes have their inputs set at the application level, and a change in parameter affects all nodes in that application. Starting with v3.0, Akumen makes use of sparse input data, making large driver models extremely efficient compared to previous versions.

• Utilizing the new “Copy page from another Driver Model” feature introduced in Akumen v3.0 will allow Driver Model libraries to be created. These can be small purpose built “approved” libraries that can be thoroughly tested (e.g. unit testing) using a dummy results set that can be deployed to other models using real live data.

• Ensure only required inputs are sent to the research grid and results.

• Make use of the Akumen Pages for those that only need scenario management, and modify inputs to give them a nicer graphical front end for entering data. They can also contain help information and data dictionaries in an easy-to-use web interface.

• Ensure apps are well documented through code comments (to ensure an easy transition for new people, and an understanding for others in the organization). Driver Model calculations also support code comments, as well as assumptions, so there is no reason why apps should not be fully documented.

• Ensure that all documentation of apps, permissions and users are up to date so that if and when an Administrator user has to complete a handover they can do so without causing too much confusion for the new admin. All documentation should be written as if the person reading it has no background in Akumen, this will make training and handover less stressful for all parties and allow new users to figure out how applications are built.