Loop Subsets of attributes and log result of each loop
Newbie
Hi,
I have a bunch of attributes / features in my dataset and want to find out which combination of features will train the best model.
Therefore I am using the LoopSubsets Operator. I specified it to "min number of attributes" = 2, so I will get at least 1 feature and the label in each loop. I use the branch operator to check, if the combination of features contains my label and then proceed in the "Then" Branch to train a model and log the performance using the "Log" operator.
The results I get don't match my training results that I get when training the model separatly on a subset of the features, by factor 10 in Absolute Error. I have the assumption that either the "Subset" Operator does something I don't expect or the log operator inside the loop doesn't work like it would outside of a loop.
How would you log the results (performance of the model or entire prediction example set) inside a loop?
This my log. I would expect absolute errors <0.1:
This is my process:
I have a bunch of attributes / features in my dataset and want to find out which combination of features will train the best model.
Therefore I am using the LoopSubsets Operator. I specified it to "min number of attributes" = 2, so I will get at least 1 feature and the label in each loop. I use the branch operator to check, if the combination of features contains my label and then proceed in the "Then" Branch to train a model and log the performance using the "Log" operator.
The results I get don't match my training results that I get when training the model separatly on a subset of the features, by factor 10 in Absolute Error. I have the assumption that either the "Subset" Operator does something I don't expect or the log operator inside the loop doesn't work like it would outside of a loop.
How would you log the results (performance of the model or entire prediction example set) inside a loop?
This my log. I would expect absolute errors <0.1:
This is my process:
<参数键= " include_special_attributes" value="false"/>
<参数键= " include_special_attributes" value="false"/>
<参数键= value =“skip_undefined_labels真的"/>
Tagged:
0
Best Answer
-
MartinLiebig
Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University ProfessorPosts:3,287
RM Data Scientist
Hi,你是否检查了自动功能工程啊perator? That should be very close of what you want to do.
BR,Martin
- Head of Data Science Services at RapidMiner -
Dortmund, Germany0
RM Data Scientist
Answers
Dortmund, Germany