Today just going to focus on writing, clarifying where I’m at, what I have so far.
I don’t think I can delve too much into ‘methodology’ yet, theres no uncertainty model yet.
I think an important thing to address is why is an uncertainty analysis useful in a decision making setting.
Refining problem (Polya)
Is it a useful mental model to think of the LCA as providing a ‘unit of information’ for further decision making, I’m not fully sure what I mean here, I would need to understanding how information flows in decision trees but my thinking that its also useful in guiding what the model (Python one here) can do? Or what its capabilities should be.
An uncertainty analysis defines information quality of an LCA. The goal then of our program is to define this quality as accurately as possible for any given scenario.
How does this relate to the value of information from decision theory? Is it relevant.
To really be influenced by what I’ve been reading, using the term quality implies something inherent in the model. An analysis here is a sense perception of that inherent feature of the model.
I think in line with this, where the model has some inherent quality, its also relevant to see its structure as best we can. As another dimension of axis of quality of information that the model provides. By this structure I mainly mean assumptions and general transparency. I think this is that overlap with the model design, searching for that correct level of transparency.
I’ve said from the start that I see two overaching problems, a design problem and an uncertainty analysis problem. They can kind of both be branched under this quality term. Interestingly, a synonym for quality is parameter, which aligns with the definition from the standards doc.
Unceratinty analaysis prpoblem referes to this quality of its output and generally pushing towards improving that quality. Where quality is a measure of accuracy. Atomicity of that chunk of information. How close it is to a point estimate.
Design problem tries to get familiar with the structure of the model, through probing it, revealing its assumptions.
I’m trying to get this overarching idea to fit somehow, so that all I do can be correlated with it. It’s influenced by N and T when they talked about vague general statements influencing design. I’m not really too sure if this is useful.
A sensitivity analysis is a probe, a stimulus response process for the model. Probing to see the models structure.
I think having this general idea of determining the models quality, strucutre
The code
How do I explain why the Python code is useful in my rolling brief?