Apply model from tree to predict the future
JanBuerger
New Altair Community Member
I followed along the "Elaborate Your Time Series Analysis" tutorial on Youtube for creating trees (like Random Forest).
I have three results, the "Performance" (something like 8%), the "Cross Validation" (which shows the predictions, but ends with the end of my input data) and at lot of trees itself.
But I am missing the main point: How and where can I predict the future? I am looking for a chart like I get in the "Cross Validation" - but for XX days in the future. Using ARIMA I just connect the forecast with "Apply Forecast", but how can I do something similar with trees?
Thanks a lot for your help!
I have three results, the "Performance" (something like 8%), the "Cross Validation" (which shows the predictions, but ends with the end of my input data) and at lot of trees itself.
But I am missing the main point: How and where can I predict the future? I am looking for a chart like I get in the "Cross Validation" - but for XX days in the future. Using ARIMA I just connect the forecast with "Apply Forecast", but how can I do something similar with trees?
Thanks a lot for your help!
Tagged:
0
Best Answer
-
Hi @JanBuerger ,I think you want to check the operators windowing and process window. They have an option to create one or more horizons (i.e. the value of the time series XX days ahead). you can use this as a label to learn a model. If you want to forecast for more than a single day (lets say 10-14 days), you can use a larger horizon and then use the Multi-Label Modeling operator.Best,Martin-1
Answers
-
Hi @JanBuerger ,I think you want to check the operators windowing and process window. They have an option to create one or more horizons (i.e. the value of the time series XX days ahead). you can use this as a label to learn a model. If you want to forecast for more than a single day (lets say 10-14 days), you can use a larger horizon and then use the Multi-Label Modeling operator.Best,Martin-1