We should have jsonschema "schema" files (which themselves are just JSON) for:
the input data (e.g. the timeseries of presentations x population strata)
the model parameters (e.g. capacities, stay durations)
the outputs (e.g. the timeseries for the daily occupation of the different resources, plus stuff about overflows, and maybe some summary statistics (although those could be derived from the raw output data))
@Markopolo141 I notice that you added src/patientpaths/input2.json in the draft_json_input1 branch. If that's just a data file, the idea is that it would validate with the abovementioned schema files.
Then we could use this to generate automated tests, and also get e.g. Rob/Katie to help us put even stricter & more accurate bounds on the input values.
Again, @Zac-HD this is probably something where @Markopolo141 can fill out the schema once there's some scaffolding in place, but it'd be useful if you set things up in a way which is most natural (as per #6 ).
We should have jsonschema "schema" files (which themselves are just JSON) for:
@Markopolo141 I notice that you added
src/patientpaths/input2.json
in thedraft_json_input1
branch. If that's just a data file, the idea is that it would validate with the abovementioned schema files.Then we could use this to generate automated tests, and also get e.g. Rob/Katie to help us put even stricter & more accurate bounds on the input values.
Again, @Zac-HD this is probably something where @Markopolo141 can fill out the schema once there's some scaffolding in place, but it'd be useful if you set things up in a way which is most natural (as per #6 ).