The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here

Explain Predictions - Missing Connection

SkyTraderSkyTrader Member Posts: 88 Contributor II
in Help
Hi there, can anyone figure out how to get the Explain Predictions operator to work? It keeps saying I have a missing connection?

Pls see images.



I'm also not sure why I can't use "Explain Predictions" as a tag?

Cheers, appreciate any thoughts on this.
Tagged:

Best Answers

Answers

  • hbajpaihbajpai Member Posts: 102 Unicorn
    edited August 2020
    Hey @SkyTrader

    The explain prediction operator does not support .forecast model objects. You might need to replace the ARIMA model with other operators that produce .model objects to utililze explain predictions operator. 
    Best,
    Harshit
  • SkyTraderSkyTrader Member Posts: 88 Contributor II

    Thanks very much @hbajpai,
    I want to find out which attributes are the main ones that are predicting my label (closing prices). 

    Can I not use ARIMA to find out which of my 10 technical indicators are the best predictors of the Dow Jones closing prices? (I mean is there nothing I can add to ARIMA to help?)

    Do you have any suggestions as to which operators will give me a model output (I'm predicting Dow Jones Closing Prices)?

    Will Random Forests work? If so how would I physically wire the Explain operator in this set up:



    Cheers for the help!

  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    Great! Cheers @hbajpai,
    What was the original name of the training and testing operators as they don't exist when I search for them? Split validation? Cross Validation?

  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    Hi @hbajpai,
    As I couldn't use Explain Predictions with ARIMA and because I've not used Random Forests (RF) before which did work thanks to your help, could you please explain the seemingly "random" prediction dates sequence chosen/presented by the RF algo?

    The dates seem to follow no pattern although my Excel Dow Jones data for the open/high/low/close was daily? Pls see image:


    Also how do I get a RF to make horizon predictions like 30 days ahead which I know is possible with RF. based on research papers I've read?

    Here is the working operator layout for anyone interested: 



    Cheers again for your help,

  • hbajpaihbajpai Member Posts: 102 Unicorn
    Hey @SkyTrader

    You can use the RF model inside the Forest Validation operator or Multi Horizon forecast operator depending on your use case. The recent webinar on Industrial time series analysis talks in detail about utilizing multi-horizon forecast.
    Best,
    Harshit
  • SkyTraderSkyTrader Member Posts: 88 Contributor II
    Hey @hbajpai
    Thanks a lot for the link I will check it out.
Sign In or Register to comment.