New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Can I use TimeSeries prediction with Vec<R4, N> input data? #1818

Open
NektoDron opened this Issue Dec 4, 2018 · 5 comments

Comments

7 participants
@NektoDron

NektoDron commented Dec 4, 2018

Is it possible in the future?
Now it makes an exception "The feature column has type 'Vec<R4, 3>', but must be a float."

@codemzs

This comment has been minimized.

Member

codemzs commented Dec 4, 2018

Are you wanting multi variate time series?

@NektoDron

This comment has been minimized.

NektoDron commented Dec 4, 2018

Yes. Also I wait for time series with next value prediction.

@singlis

This comment has been minimized.

Member

singlis commented Dec 4, 2018

Hello-
Thank you for bringing this up. The multivariate time series is not supported at this time, however it is something that we are tracking here:
#1696

@Zruty0

This comment has been minimized.

Member

Zruty0 commented Dec 6, 2018

@NektoDron , predicting the next value is not something we have on the immediate plans. However, you can simulate these 'auto-regression' models by taking a sliding window of the original time series and training a regression model to predict the next value in the series.

Of course care should be taken to make sure seasonality and trends are captured properly, but I think this is a well-known area of research.

@NektoDron

This comment has been minimized.

NektoDron commented Dec 6, 2018

@Zruty0 , I've tried to do this. But it's too slow, because I should always retrain many models to simulate how prediction works on the history data.

@sfilipi sfilipi added the enhancement label Dec 7, 2018

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment