Menu
▾
▴
#222 Units for model parameters - should be either disabled or made fully functional
Currently the user can supply units for model parameters like lag, Y0 or Ymax. This information is highly useful, e.g. especially in the case of the Init concentration N0 (and probably also in case of lag). Example: in the Predictor View currently it is not clear in which unit the initial concentration should be entered (this field should by synchronized with the selected y-unit).
Currently the units, that a user specifies for parameters that are not dependent or independent variables, get lost - but the user does not notice this. So ideally we would aösp use these units if provided (e.g. for Y0). If this is to cumbersome now we should make the unit configuration fields of formula parameters dysfunctional and empty (by this we could then also remove the warning of the model write, that a unit is missing for those parameters).
When the user specifies units for parameters, these might be inconsistent with the variable units. Therefore I am in favor of not allowing the user to specify parameter units.
@Armin: If you agree, could you disable the unit field for parameters in Formula/Model Creator?
Will I still be able to assign units to model formulas which I want to save in the DB? Once I had saved a model with the wrong time unit and wondered why the half-life was so fast. Without unit I would not have found the mistake. And knowing which unit A0 has is necessary to assign data in the correct format because otherwise the formula computes results which do not make sense (biologically).
I am fine with this provided that there is no inconsistency or lack of information when calculating things. Chris, you are the person who should judge this. Please be aware of Alex' point. My impression is that we need units for all parameters...
Right now the parameter units could only be used to document parameters (as Alex mentioned). The user can also do this by writing a description text. Therefore we should remove parameter units from Model/Formula Creator.
Or we could make it obvious to the user, that parameter units are only descriptions.
@Chris: Would it be a big issue to extend our XML-based data format that we use between our nodes such, that information on units for model parameters are just transferred if provided?
I think it would be great if this could be accomplished. The implementation of features that make use of this unit information might then be a different issue and maybe one can implement that for some nodes later. I'm really afraid, that a "description" does not help, as e.g. in the predictor view you do not see the description of Y0 (but you see the missing unit). Would that be an idea?
There is already a parameter unit in the XML-based format. I just do not use it anywhere. But I think it is written in the DB. I'm fine with it, if people are using it for documentation.
The unit for the starting value "LOG10N0" is useful, but it should be the same unit as the dependent variable in all serious models. And the "lag" unit should be the same as the time unit.
Yes, this rule make sense. Do you have a suggestion how we could indicate already in the formula creator node which parameter is a "N0"-like or "lag"-like parameter and how this could be saved in the XML-format? This kind of information would also make life much easier in several other nodes (including FPL).
Yes, nice to have, but I don't see an easy way to do this -> 2.0
PMM-Lab/FPL both need to know, which parameter represents the initial concentration. Especially in FPL it is not possible to conveniently calculate the agent concentrations through a whole food process chain because at the moment the user needs to do the same assignments in every FoodProcess-Node.
Currently, the user can indicate in the FormulaCreator / ModelCreator, which parameter is independent. Would it be an option to call the column instead of "independent" perhaps "Parameter category"? With a drop-down menu the user could then assign "independent" and "initial concentration" (perhaps also lag?).