sample weights have come up a few times on github and on here but are there any supported strategies for weighting fits, both at an individual data point level, and a series level, in

MLForecast

? Being able to weight individual date values will enable modeling strategies like this (although this might equally well be handled with an exogenous variable, albeit less confidently). Series level weights would be useful for intentionally biasing the loss toward performance on important series (e.g., revenue weighted global fits in an portfolio of products, with varying sales performances). Are there any plans for this the moment?

Are you refering to this github solution?

Thats the solution I've seen, yeah

What's wrong with it?

Nothing is wrong with it at all as it does work. It's more a question about the UX as the proposed solution diverges from what is established in the rest of the Nixtla ecosystem, and it's also specific for LightGBM, if that makes sense? Support for a generic weighting function is one of the few gaps in the Nixtla libraries that I have seen supported in other forecasters

The hard part of having an argument like that is that you might provide an array with the same size as your input, but if you set dropna=True then you'll have a different number of samples on the fit step, so that'll probably cause a lot of confusion. I'll take a look at your example to see how they handle that