Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

microprediction.com

Post reply on HN
Print view View original

41–50 of 82 posts

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#41

Post by carlosf »

jd115 wrote: post #23
Fordec wrote: post #17

Earlier quoted context omitted.

Personally I find there is one important factor, and one factor alone. Context. Seasonality is the context that there's a seasonal driver at play. The context of a public holiday can explain a decrease in sales on that day. The context of a football match can explain a spike in transport demand near a stadium. The context of the presence of a heat dome predicted by pressure data can explain record temperature figures…

Is there also "one factor alone" in predicting stock market performance? Foreign-exchange rates? Lottery numbers?

Yeah models assume future data can be predicted with past data, which of course is not always true in real life due fundamental limitations of statistics, such as very fat tailed distributions in stock markets.

Competent data scientists must be able to spot those cases quickly.

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#42

Post by lr1970 »

As someone who spent a good part of my professional career in forecasting and time-series analysis I would like to point out that "point-forecasts" are mostly useless in many import practical applications such as FinTech, e-commerce, sports betting, etc. Point-forecast models such as Prophet fail to give you a meaningful measure of uncertainty of the predicted value. A much better approach are probabilistic forecasting models that predict the probability distribution of the random variable of interest [0]. Probability distribution is the right language to express prediction values and their uncertainty at the same time. And the decisions based on forecasts oftentimes take the uncertainty measures in consideration, e.g. portfolio risk optimization or buying decisions in e-commerce.

[0] https://en.wikipedia.org/wiki/Probabilistic_forecasting

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#43

Post by dcl »

lr1970 wrote: post #42

As someone who spent a good part of my professional career in forecasting and time-series analysis I would like to point out that "point-forecasts" are mostly useless in many import practical applications such as FinTech, e-commerce, sports betting, etc. Point-forecast models such as Prophet fail to give you a meaningful measure of uncertainty of the predicted value. A much better approach are probabilistic forecasti…

Do you have any favourite libraries for producing such?

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#45

Post by microprediction »

I would recommend Python fans take a look at https://www.microprediction.com/blog/popular-timeseries-pack... and the Elo ratings for timeseries (e.g.) https://microprediction.github.io/timeseries-elo-ratings/htm... The timemachines package wraps prophet and others for the purpose of comparision https://github.com/microprediction/timemachines

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#47

Post by microprediction »

dcl wrote: post #43
lr1970 wrote: post #42

As someone who spent a good part of my professional career in forecasting and time-series analysis I would like to point out that "point-forecasts" are mostly useless in many import practical applications such as FinTech, e-commerce, sports betting, etc. Point-forecast models such as Prophet fail to give you a meaningful measure of uncertainty of the predicted value. A much better approach are probabilistic forecasti…

Do you have any favourite libraries for producing such?

https://www.microprediction.com/blog/popular-timeseries-pack...

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#48

Post by microprediction »

em500 wrote: post #28

I'm a professional forecaster (i.e. getting paid for it) at a large e-commerce company. We have extensive experience with Prophet and a host of other approaches (all the traditional models in Hyndman's book/R package, some scattered LSTM/NN implementations). Here's my quick take (the article is a lot more extensive than the median blogpost, and likely warrants a more extensive study than I have time for right now.) P…

Take a dataset fit by prophet. Then run the residuals through timemachines. You'll see where I'm coming from.

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#49

Post by glial »

lr1970 wrote: post #42

As someone who spent a good part of my professional career in forecasting and time-series analysis I would like to point out that "point-forecasts" are mostly useless in many import practical applications such as FinTech, e-commerce, sports betting, etc. Point-forecast models such as Prophet fail to give you a meaningful measure of uncertainty of the predicted value. A much better approach are probabilistic forecasti…

Can you elaborate a bit? Prophet can use MCMC sampling and includes uncertainty in its forecasts.

Re: Is Facebook's “Prophet” the time-series Messiah or just a naughty boy?

#50

Post by microprediction »

sambucini wrote: post #24

i've used prophet along many other ts methods for price forecasting in energy trading. my experience is that prophet is ok, but rather opaque. having tried many packages, I've always come back to the classical statistical methods seeing benefits in transparency (what's actually going on? impact of regressors?), speed and most importantly that these methods force the user to think about what's happening in the data an…

Can't say I was surprised to see TSA near the top of the rankings https://microprediction.github.io/timeseries-elo-ratings/htm...
Post reply on HN