[Scikit-learn-general] Combine criterions for building a tree

Suppose I want to create a regression tree accepting both continuous linear data and circular data. If I implement a new RegressionCriterion specific for circular data, how difficult would it be to grow a tree combining to different Criterions (ie MSE and the new CircularCriterion)? I suppose the

Re: [Scikit-learn-general] Combine criterions for building a tree

2014-01-29 Pablo Rozas Larraondo p.rozas.larrao...@gmail.com: Suppose I want to create a regression tree accepting both continuous linear data and circular data. If I implement a new RegressionCriterion specific for circular data, how difficult would it be to grow a tree combining to different

Re: [Scikit-learn-general] Combine criterions for building a tree

Hi Pablo, I am not sure re-implementing a new criterion is what you are looking for. Criteria are made to evaluate the goodness of a split (i.e., a binary partition of the samples in the current node) in terms of impurity with regards to the output variable - not the inputs. What you should do

Re: [Scikit-learn-general] Combine criterions for building a tree

Hi Gilles, Thanks for your help, you're right, what I'm looking for is a new implementation of a Splitter that deals with circular data. I suppose, I will have to test all my variables at every node to find the optimum split measured with a criteria. What it's still not clear to me is if it

Re: [Scikit-learn-general] Combine criterions for building a tree

I suppose, I will have to test all my variables at every node to find the optimum split measured with a criteria. What it's still not clear to me is if it exists an elegant way of choosing the right splitter depending on the variable, via tagging or any other solution. Yes, we don't have any

Re: [Scikit-learn-general] Combine criterions for building a tree

Just to clarify things Pablo: are your circular variables input features or target regression variables in a multi-output regression task? -- Olivier -- WatchGuard Dimension instantly turns raw network data into

Re: [Scikit-learn-general] Combine criterions for building a tree

Thanks Gilles and Lars, I will try to figure out what is the best way to hack the current implementation of the tree builder to combine different splitters. Olivier: What I want to implement is a tree that can combine both linear and circular variables as input features and also the target

Re: [Scikit-learn-general] Combine criterions for building a tree

Alright thanks for the clarification. In that case you will to hack a lot indeed, good luck :) -- WatchGuard Dimension instantly turns raw network data into actionable security intelligence. It gives you real-time visual

Re: [Scikit-learn-general] Combine criterions for building a tree

I think you're right, it might be quite a big of a hack... :-) Anyway, if any of the main developers think it might be worth enabling this kind of extensibility (handle different variables types) to the tree implementation I'd be more than happy to contribute under limited direction. Otherwise, I

Re: [Scikit-learn-general] Combine criterions for building a tree

a hack that might be good enough could be to use as features the cos and sin of the angle and split the output (using then multioutput trees) Alex -- WatchGuard Dimension instantly turns raw network data into actionable