Options
Applying new attributes created by Yagga
Maven
Hello,
I have a rapid miner training process which uses the Yagga operator and it creates two additional attributes it calls gensym20 and gensym21. The example set with these new attributes along with a few of the original ones goes to a SVM operator and creates a model which is written to a file via the Write Model operator.
I would like to use this model in a second rapid miner process (I call a reader). I load new data and want to generate the attributes gensym20 and gensym21 without retraining the system.
The Yagga operator has three output ports: example set out, attribute weights out and performance out. I connected the attribute weights out port to a Write Weights operator. This only tells me that the gensym20 and gensym21 are used but not how to calculate them.
How do I calculate the gensym20 and gensym21 attributes with new data, without using the Yagga operator?
Thanks in advance,
Cleo
PS I am using RM 5.0 RC
I have a rapid miner training process which uses the Yagga operator and it creates two additional attributes it calls gensym20 and gensym21. The example set with these new attributes along with a few of the original ones goes to a SVM operator and creates a model which is written to a file via the Write Model operator.
I would like to use this model in a second rapid miner process (I call a reader). I load new data and want to generate the attributes gensym20 and gensym21 without retraining the system.
The Yagga operator has three output ports: example set out, attribute weights out and performance out. I connected the attribute weights out port to a Write Weights operator. This only tells me that the gensym20 and gensym21 are used but not how to calculate them.
How do I calculate the gensym20 and gensym21 attributes with new data, without using the Yagga operator?
Thanks in advance,
Cleo
PS I am using RM 5.0 RC
0
Answers
you could use the Write Constructions operator to write the constructions of an example set to file and the respecting read operator to reconstruct them on an apply set.
In future this will be changed to the much better concept of preprocessing models, but we didn't manage this to include in 5.0
Greetings,
Sebastian
Your suggestion works perfectly.
Cheers,
Cleo