Re: [Scikit-learn-general] Learning with counts feature transform.

2015-09-18 Thread Christos F. Papadopoulos
Hi Andy, In this competition, it helped me get to 1st place for a bit, and the current 1st place holder is also using this. https://www.kaggle.com/c/sf-crime/forums/t/15836/predicting-crime-categories-with-address-featurization-and-neural-nets I am pretty sure that the data is public for this spe

Re: [Scikit-learn-general] Learning with counts feature transform.

2015-09-18 Thread Andy
Hi. I think it would be great to add this transformation, it has been widely advertised by Owen. It would be great to have some public datasets (well-established preferable) to demonstrate the usefulness. The sf-crime is public, so that would be a good first step. Can you provide a full-fledged

[Scikit-learn-general] Learning with counts feature transform.

2015-09-17 Thread Christos F. Papadopoulos
Hi all! Are there any plans to implement something like a countfeaturizer? https://msdn.microsoft.com/en-us/library/azure/dn913056.aspx It is a generic method for feature extraction that encodes multi-valued features based on classification count data. I've found it very useful, for example on th