[chroem-]

The point of Prophet is to make time series accessible to people who aren't experts in time series. Yes, it has some well known failure modes, but using Prophet always beats a static LY "forecast." This article is like complaining a Nissan Leaf can't outrace a Tesla.

[akdor1154]

Does it though? The article explicitly shows cases where simple last-value performs better.

[chroem-]

Yes, it does. I'm not sure what the author is trying to prove, but he's feeding a bunch of non-seasonal data into a highly seasonal modeling technique, and then hurling insults when it doesn't work because that's not what it was originally designed for. I'm sorry, but it just seems childish.

[zwaps]

But this seems to be a bit contradictory. Both facebook and you claim that prophet is a package to easily do better than other simple approaches, thus providing at least a good baseline for applied data people.

By contrast, the article suggests that prophet does badly in many common forecasting situations. So while prophet may do fine with piecewise linear seasonal data, it apparently is far from a good default.

There’s an inconsistency in the message here. Is prophet a tool for a very specific type of time series? Or, as the marketing claims go, is it a general easy to use default?

Much like facebook, you made both claims successively, which is I think the point of the article.

[chroem-]

The vast majority of time series data that you would encounter in a business setting is seasonal. Most of the datasets he is using are tiny and they haven't had enough history for the seasonality to develop.