Community
Participate
Working Groups
I think it is essential to integrate a feature that generates on request automatically a report on the scenario where all model assumptions, parameters, mathematical formulas used and also the data used are documented. Ideally this documentation should be formatted in two ways: First: a human readable format, e.g. one could structure it in analogy to the scenario hierarchy; Second: a XML-format, such that one can automatically restore the complete scenario from that file and other could use that file to import all details into their software (of course this is a long time goal). In the long term perspective one could additionally extend this documentation on a sensitivity analysis where the user can estimate which assumption or which model parameter has the biggest effect on a output parameter relevant for the user.
Matthias, I'm thinking we can use the dublin core for this. We would need several steps. 1) We need to autogenerate more information on assumptions used in the disease models (expressed in the dublin core) 2) Users need to fill in scenrio specific info as they compose a scenario. 3) We could then provide a tool that summarized all dublin core content from any particular scenario. Do you think this would satisfy the feature request?
Feature request
Jamie, I think the dublin core might be the right choise. What I'm a bit concerned about is the target milestone. I think 1.1.2 is too ambitious, 1.2 would be OK as well. Additionally I would suggest to split this bug into two: this one will in future cover the human readable report generator; the new one covers the XML-language for epidemiological models and scenarios.
Matthias, The target is set by default to 1.1.2 right now so this is just a mistake. I agree this is 1.2 or even later as it's a big task and we need to get it right. I also agree it's too requests 1) Human Readable Dublin Core Report Generator: Summarizes all metadata in a scenario 2) Better/more detailed autogeneration of Dublin core for disease and population models. btw I think this is also related to a research effort we will start to derive aggregate estimates of uncertainty based on component undertainty in each dublin core element. Should be part of the full summary.
Working on this. Will update progress on this weeks call. Question: Should the html report page contain javascript?
Report generator is implemented so I'm resolving as fixed. The view may move with Matt's new logging framework but the reports are now in place and released with 1.2.1
