- Today I need to work on cleaning up the code, or pushing towards a ‘stable’ version of the code for the coming weeks.
- Just having that flexible underlying representation.
- The underlying code confuses me at the moment.
- What makes sense would be to have a stable version and then on my machine I could have a little helper file (with helper functions) if they become more and more useful I can add them to the repo.
- It seems we’re stuck between just churning out an analysis and a bit of rigour. It seems that we should take our time with each unit process, but this would mean that we expect this piece of software to be used when I leave which I believe is highly unlikely.
- Would there be a way to mark the uncertain numbers. Should just try and get a stable version first.
- Considering this done for the time being before I shoot myself formatting Excel cells.
- The issue is that I’m using the direct numbers in Excel cells, not their formatted version.
Current thoughts then:
- Stop developing it as a tool, as I just don’t think it’ll be used this way.
- This means just plugging in numbers. So I’m just going to run it in Python.
- Start to focus more on propagation using MC.
- In terms of structure of report, how much do I explore topics like open science, value of information, model building vs just tick off the boxes of uncertainty analysis?
- Start trying to produce results.
- Would need to research more about parameters taken as ‘certain’.
- Can we settle on a stable version (as in replication of n processes).
Just trying to lean back into the uncertainty aspects which are interesting. the measureJK
AM on Lunar podcast. We should think more about design of a thing, an issue with the start up mindset. Also, if you’re setting up your personal blog, what is the goal of your blog? this should guide decisions.
I think about this in relation to my current work. I’m trying to just hash something out for its own sake without proper thought of what we want it to do. What do I want the code to be able to run, that should be guiding it. At the moment its a transitory period. Moving from designing it for use by others as a replication of LCA tool to a purely (possibly single use) uncertainty analysis tool. That’s totally fine, would have been nice to figure it out sooner but still. It’s an uncertainty analysis tool then, what should it be able to do?
What happens if an uncertainty analysis completely invalidates results?