Soundwave comments

Results 7 comments of


                                            Soundwave

question or to discuss: how to find optimal parameters of xgb?

Early stopping param is not supported in scikit-learn wrapper yet... maybe we can ask whether he want to support those params in scikit-learn wrapper. :)

question or to discuss: how to find optimal parameters of xgb?

Parameters : - eta : step size shrinkage used in update to prevents overfitting. - alpha / lambda : L1/L2 regularization term on weights. - subsample : subsample ratio of...

question or to discuss: how to find optimal parameters of xgb?

FYI https://github.com/dmlc/xgboost/blob/master/doc/how_to/param_tuning.md Most time default parameters works pretty well.

to discuss: Kaggle tips and tricks - someone lightening talk?

From @jeongyoonlee, ensemble a xgb and a deep learning model. :)

question or to discuss: sensitivity to initial conditions especially in assigning default branch directions for missing data

DMatrix can specify which number is missing value.

question: future plan of xgboost

Some refactor. Make things more moduler including plug in system. Possible JS version. (porting to mobile?) Fast prediction. (Specialized prediction API.) Ingest from dataframe directly.

to discuss: main points of xgboost paper on arxiv

4.1 Column Block (data structure for parallel learning)