https://github.com/nixtla logo
m

Mark

07/19/2023, 7:53 PM
@Kevin Kho @Max (Nixtla) I assumed optimalcombination reconciliation was going to be the most intensive, i guess maybe i was wrong and it was probably middleout? I haven't narrowed it down yet, but it worked running it just on the single driver node 😎
k

Kevin Kho

07/19/2023, 7:56 PM
i dont have a sense for computational cost for each, but just wondering, did BottomUp or TopDown finish?
m

Mark

07/19/2023, 7:56 PM
bottomup runs instantly, didnt run an individual topdown just yet, im assuming itll be about the same as bottomup
k

Kevin Kho

07/19/2023, 7:59 PM
ah gotcha
m

Max (Nixtla)

07/19/2023, 8:34 PM
You are going to love this great feature by @fede (nixtla) (they/them) and mcsqr: https://github.com/Nixtla/hierarchicalforecast/pull/210
m

Mark

07/20/2023, 11:29 AM
I'll add this method @Max (Nixtla) and provide feedback