Closed avouspierre closed 1 month ago
From my understanding, it looks like this PR does two things:
I'm not saying this is a bad idea, I just don't know the reasoning behind it. Looks like iAPS has been using 10-days for over a year now.
Currently, Trio populates an array with the TDD value collected every loop cycle from the past two hours and then averages them together. Each of these values is the total insulin dosed in the past 24 hours. This means it includes up to 24 values with insulin data from the past 26 hours
This PR, however, would increase the size of this array to include the TDD value from every loop from the past 24 hours instead, which would include *up to 288 values with insulin data from the past 48 hours.
I‘m genuinely confused. This was brought up as a topic of discussion and is now PR‘d with any further conversation about it or its side effects.
@dnzxy Sorry if the PR is too quickly. I noted the PR interested more than discussion in GitHub ;-)
@MikePlante1 :
I close the PR and update the discussion.
Oh, please don’t close, let’s just make it a draft for now. You did the work already, let’s not waste it 😊 Is that okay with you? I‘ll re-open and change to draft, if so 🤔
I could be wrong but I think I recall Tim Street commenting somewhere that in AAPS they found two hours unsafe.
Oh, please don’t close, let’s just make it a draft for now. You did the work already, let’s not waste it 😊 Is that okay with you? I‘ll re-open and change to draft, if so 🤔
Don't worry. it is more the argument of Mike than your remark 😂
Modify weighted_average to respect the 24 hours of data and 14 days of total TDD.