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

Process Window: Include more than one attribute

tomMEMtomMEM Member Posts: 15 Contributor II
Hello, I try to prepare a collection of exampleSets for LSTM deep learning. Using Process Window it is possible to get easily the collection for one time-series attribute (temperature) as well as a horizontal label (temperature + 5). However, I could not find out how to get included several additional time-series attributes (humidity, cloud level etc.). I can do it one by one using join, but it is not a very flexible approach.
Suggestions are welcome. Best. T

Best Answer

  • ceaperezceaperez Member Posts: 541 Unicorn
    Solution Accepted
    Hi @behnish,

    the solution for recurrent operations is always the loop operators.
    If you need to create an exampleset for a subset of your attributes you can run a loop attribute and inside of it a windowing operator using %{loop_attribute} as target attribute.
    It will operate over your attributes creating the same windows size and horizon size for each one.  

    By other hand, if you only need to create one horizon attribute and include the any other attributes into the exampleset you need to select a subset option into the attribute filter type in windowing options and then select all the attributes that you need

    Best 

    Cesar

Answers

  • tomMEMtomMEM Member Posts: 15 Contributor II
    ceaperez, thank you for the response. I installed the 9.10 RM and replaced the old Process Windows with the current one. The old one's compatibility stopped at 8.2. The new Process Windows allows to set an attribute subset. For additional flexibility I will also try in parallel your LOOP suggestion. Thank u. T
Sign In or Register to comment.