xgboost-adv-workshop-LA

xgboost-adv-workshop-LA copied to clipboard

question: distribution of scores

2

Weird (uncalibrated?) distribution of scores (though it does not matter for AUC): https://github.com/szilard/xgboost-adv-workshop-LA/blob/master/score_distrib/distrib.R

szilard

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

5

I usually do a grid search of the following parameter set: param_grid = { "learning_rate" : [ 0.1, 0.05, 0.01], "n_estimators" : [ 100, 500, 1000], "max_depth" : [ 4,...

lihang00

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

1

szilard

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

1

e.g. is it better to assign all to left-direction/right-direction/random?

ikukuyeva

question: future plan of xgboost

1

Current version of xgb is pretty mature and can be used widely in different scenarios. What is the future plan of xgboost?

lihang00

I'm working on a generalization of PCA to high-dimensional large, dense data sets (e.g. latitude, longitude, values -- and ideally time), where all of the cross-relationships are important. Do you...

ikukuyeva

ikukuyeva

ikukuyeva

It seems to me that the train method and the predict method expects the train and test sets to have the same ordering of the columns, am I correct? If...

damienbenveniste

The implementation of Random Forest through Xgboost is not truly Random Forest as defined by Breiman, am I correct? In the Xgboost case the features seems to be sampled at...

damienbenveniste