Great presentation as always! Any timeline for a public BETA, soon?
@Nixtla
4 ай бұрын
Thanks! You can now get an api key with free trial at dashboard.nixtla.io
@carrocesta
5 ай бұрын
I have just tried it and it is much simpler and faster than keras for univariate LSTM, very cool!
@Nixtla
4 ай бұрын
Thanks! We are always happy to hear feedback.
@mihairadulescu5541
2 ай бұрын
Guys you just enabled new products and use cases!!! Amaizing work
@FindMultiBagger
11 ай бұрын
what about data privacy if you are using gpt ?
@MingkaiLiu-j9q
2 ай бұрын
You just put many data scientists out of job ;D
@pietropeterlongo2695
Жыл бұрын
"it is also, clearly, a marketing stunt!" apart from the good joke, I really like how Nixtla keeps a ground to earth communication that acknowledges but does not unnecessarily ride the hype!
@Nixtla
4 ай бұрын
Thanks for the support :)
@Supernumerary
8 ай бұрын
Influences that were previously predictable can suddenly change. Example, there’s plenty of trouble in the Middle East, yet crude oil prices have not increased.
@Nixtla
4 ай бұрын
That is correct indeed. One common way of addressing the effect of other factors in your target variable is by including exogenous variables. Here is a tutorial: docs.nixtla.io/docs/tutorials-exogenous_variables
@lashlarue7924
11 ай бұрын
I'm using Uber's Orbit library and a gradient boosters. Would love to try this...
@Nixtla
4 ай бұрын
You can now get an api key with free trial at dashboard.nixtla.io
@RBLevin
11 ай бұрын
Super boring.
@Nixtla
4 ай бұрын
Thanks, we will try to make Max's next presentation more exciting.
@phlippseitz
Жыл бұрын
Looks very promising, would love to try it
@Nixtla
4 ай бұрын
You can now get an api key with free trial at dashboard.nixtla.io
@omarfarooq5772
11 ай бұрын
When API
@Nixtla
4 ай бұрын
You can now get an api key with free trial at dashboard.nixtla.io
@abdolreza82
7 ай бұрын
How are you going to train your model to forecast chaotic time series? Like Lorenz attractor.
@Nixtla
4 ай бұрын
There are certain series that are indeed very hard to forecast without additional information. Chaotic processes are particularly challenging to predict. The premise of TimeGPT (and other time series models) is that the past, combined with additional relevant information, can be helpful in predicting the future.
Пікірлер: 21