CalTRACK 2.0 Standing Meeting - Shared screen with speaker view
hi phil, could you please explain why bootstrapping helps you get to the ground-truth? thanks
Also, please explain again why the figures are so far below 100%…?
How are potential elevation differences between site and weather stations handled?
i can follow up with you offline phil. sorry, my mic on my computer isn't working, hard to talk this through just via chat
Can you speak to the data quality of the different sites and how that factors into an algorith to provide robust temperature data.
Follow on from Steve Schmidt's question - can you articulate what "56%" denotes exactly. I'm still trying to get my head around this metric
Perhaps the results are senstive to the particular ratio of weather dependent load versus weather indepenendent loads for this site
Question on the weather dependency: results appear to show low sensitivity to using non-local weather for training and prediction period. How about the impact if you want to normalize results to local TMY data? Might be worth some follow up
How do you handle cases where the balance point is on the extreme of the search range?
I am a little confused about the conclusions and recommendations of the baseline period test to determine impact of baseline measurement period on normalized baseline energy use. Is there more detail on your analysis than what is provided in Github?
Have you looked at the correlation between values bunched at the extremes of the balance point ranges and loads with little to no temperature sensitivity.
By "R-square" you do you mean "adjusted R-square"?
As a P4P program implementer, we do not want any homes “disqualified” by CalTRACK. I think that’s pretty obvious.