time series forecasting recipe has bugs?

R_zen
R_zen Registered Posts: 1

I'm trying to use time series forecasting recipe to predict multi-time series data.

But it's always stopped at resampling steps in time series forecasting recipe as showed below.

To debug, I created dummy data attached below with my time sequence which is same as my original data . Q is sequence number (1,2,3,4....), Q1 is doubled Q . It was stopped as my original data in resampling.

Even I used time sereis preparation plugins successfully to resample the dummy data, finnally it was failed in resampling steps of time series forecasting recipe.

I attached the log and settings as well.

Did anyone meet same problems?


Operating system used: Win 10

Answers

  • NicolasD
    NicolasD Dataiker, Dataiku DSS Core Designer, Registered Posts: 12 Dataiker

    Hello R_zen,

    You're trying to train a timeserie per value in Q, however for the identifier "2" there's a single data point, so DSS is not able to train a time serie for it.

    If you really want to train a timeserie per value in Q, you'll have to curate your data so that all identifiers contain at least 2 data points.

    I hope this helps Nicolas

Setup Info
    Tags
      Help me…