How to give feature weight in xgboost?

[matt]

I am dealing with multi-view features. Each view feature has different distribution and feature space scale vary as well. I want to apply weight to each view. How can I do this? I prefer to code in python ? does this require changing exiting source code? if yes, where? and I would like to read any document regarding how code is organized?

Thanks
Matt

[hcho3]

There are outstanding feature requests:

• Ability to define probability per feature when using feature sub-sampling: https://github.com/dmlc/xgboost/issues/3754
• Custom (user-defined) split evaluator: https://github.com/dmlc/xgboost/issues/4230