I have several weights in my tree that I want to apply. SetSignalWeightExpression take a parameter a single branch value. Is there a way (or another method) to do something like SetSignalWeightExpression (“weight1weight2…*weightn”)?
The weight expression should take more than one Tree branch as it does DataLoader::AddVariable
- ok, thank you.
- I saw from your talk today in IML that there is an automatic way to optimize parameters “Hyper-parameter optimisation”, is that correct? if yes how can I do that for DNN? Is it available only for BDT?
At the moment this is not yet available for the DNN. It is presently not very practicable in that case because it will require training every time a new DNN, which is very time consuming
I have a weight w, and a boolean variable v in my tree. I also have a number n which is not in the tree.
I want to apply a total weight to each event: w * n * (v<1).
Is it possible to do that with “SetSignalWeightExpression” this way:
SetSignalWeightExpression( " w * n * (v<1)" );
Indeed as n is not a variable in the tree, I not sure I can’t do that. But if indeed I can’t, is it still possible to do:
SetSignalWeightExpression( " w * (v<1)" );