Option to configure data preparation Auto Model timeout
In the last step of Auto Model, i can choose the option "Automatic Feature Selection" with "Automatic Feature Generation", although i set 60 minutes or more, the process ends earlier.
My question is, where can i configure this value to prevent the process ends?
Version: RapidMiner Studio 9.3.001 (rev: 289b20, platform Win64)
My question is, where can i configure this value to prevent the process ends?
Version: RapidMiner Studio 9.3.001 (rev: 289b20, platform Win64)
Find more posts tagged with
Sort by:
1 - 5 of
51
I would like to squeeze the maximum computing power of one of the computers we have and see if the result differs.
Or am I wrong and doesn't the hardware matter in this case?
Thanks
Or am I wrong and doesn't the hardware matter in this case?
Thanks
Hello @User222697
Do you mean by setting different time limits? It might if you have a process that takes so long and the optimization is being stopped because of the time limit. They also differ a bit if you change the accuracy or complexity in the automatic feature engineering operator.
Do you mean by setting different time limits? It might if you have a process that takes so long and the optimization is being stopped because of the time limit. They also differ a bit if you change the accuracy or complexity in the automatic feature engineering operator.
Hello Varunm1,
Do you mean by setting different time limits?
- Exactly, how can i change the value?
Thanks for your fast response
Do you mean by setting different time limits?
- Exactly, how can i change the value?
Thanks for your fast response
The way you changed earlier is correct but as I said, it is the maximum time limit and if your process and data doesn't need that much time it will complete early and you cannot test that time. Try setting time to 10 seconds and 20 seconds to observe if there is any change rather than setting to 60 minutes.
Sort by:
1 - 1 of
11
The way you changed earlier is correct but as I said, it is the maximum time limit and if your process and data doesn't need that much time it will complete early and you cannot test that time. Try setting time to 10 seconds and 20 seconds to observe if there is any change rather than setting to 60 minutes.
Is there any specific reason you want to run the feature engineering for 60 or more minutes? The values you input (60 minutes) is the limit which is a maximum time that it's run, but if it finds the best subset and generates features in less time it will be completed earlier than that time.