[HN Gopher] Time Series Forecasting with Graph Transformers
___________________________________________________________________
Time Series Forecasting with Graph Transformers
Author : turntable_pride
Score : 58 points
Date : 2025-06-17 18:05 UTC (4 hours ago)
(HTM) web link (kumo.ai)
(TXT) w3m dump (kumo.ai)
| ziofill wrote:
| I can't stand websites that override scrolling
| pealco wrote:
| Most of my time interacting with this site was spent in
| developer tools, trying to figure out where the scrolling
| behavior was coming from. (Couldn't figure it out.) I can't
| understand why people are still doing this in 2025.
| almosthere wrote:
| Most likely the developer is using a Windows computer.
| bestest wrote:
| Enter this in the console:
|
| document.body.onwheel = (e) => e.stopPropagation();
| rossant wrote:
| I came here to say this. Don't mess with my scrollbar. Ever.
| loehnsberg wrote:
| I disregard anything on time series forecasting from any entity
| that uses Facebook Prophet as a benchmark.
| esafak wrote:
| Why? That is what everybody uses. What do you use?
| loehnsberg wrote:
| L1-regularized autoregressive features, holiday dummies,
| Fourier terms (if suitable in combination) yield lower test
| errors, are faster in training, and easier to cross-validate
| than Prophet.
| esafak wrote:
| With which library though? Is it fast enough for
| production?
| hotstickyballs wrote:
| Sounds like prophet with extra steps
| melenaboija wrote:
| For such strong and personal statement I have to ask why.
| Worksheet wrote:
| If you arrived into, say, London and googled "Best fish and
| chips" would you believe that the top result gives you the
| meal that you're after?
| hotstickyballs wrote:
| Why not? It's definitely a useful benchmark
| frakt0x90 wrote:
| Prophet is great and we use it for multiple models in
| production at work. Our industry has _tons_ of weird holidays
| and seasonality and prophet handles that extremely well.
| tech_ken wrote:
| This is sales research, and after "CAGR in a GSheet" FB Prophet
| is what's going to be most recognizable to the widest base of
| customers.
|
| FWIW seems like the real value add is this relational DB model:
| https://kumo.ai/research/relational-deep-learning-rdl/ The
| time-series stuff is them just elaborating the basic model
| structure a little more to account for time-dependence
| cwmoore wrote:
| "Here, sign this." accept all cookies
| meindnoch wrote:
| 1. Stop messing with my scrolling.
|
| 2. If this really worked, you'd be making billions on the stock
| market. The fact that you don't, tells me it doesn't work.
| tech_ken wrote:
| > If this really worked, you'd be making billions on the stock
| market
|
| That's kind of a weird thing to say given that the market cap
| for quantitative finance is well over a billion dollars, and
| this product clearly seems to be targeting that sector (plus
| others) as a B2B service provider. Do you think that all those
| quantitative trading firms are using something other than time-
| series analytics?
|
| Also, setting aside the issue of whether time-series
| forecasting is valuable for stock-market trading, it seems like
| the value add of this product isn't necessarily the improved
| accuracy of the forecasts, but rather the streamlined ETL ->
| Feature Engineering -> Model Design process. For most firms
| (either in quantitative finance or elsewhere) that's the work
| of a small dedicated team of highly-trained specialists. This
| seems like it has the potential to greatly reduce the labor
| requirements for such an organization without a concomitant
| loss of product quality.
| cye131 wrote:
| I'm not a fan of this blog post as it tries to pass off a method
| that's not accepted as a good or standard time series methodology
| (graph transformers) as though it were a norm. Transformers
| perform poorly on time series, and graph deep learning performs
| poorly for tasks that don't have real behaviorial/physical edges
| (physical space/molecules/social graphs etc), so it's unclear why
| combining them would produce anything useful for "business
| applications" of time series like sales forecasting.
|
| For those interested in transformers with time series, I
| recommend reading this paper: https://arxiv.org/pdf/2205.13504.
| There is also plenty of other research showing that transformers-
| based time series models generally underperform much simpler
| alternatives like boosted trees.
|
| After looking further it seems like this startup is both trying
| to publish academic research promoting these models as well as
| selling it to businesses, which seems like a conflict of interest
| to me.
___________________________________________________________________
(page generated 2025-06-17 23:00 UTC)