Xgboost: Unable to pass attribute "booster" to XGBClassifier (python)

[ikamensh]

Environment info

Operating System:
Ubuntu

xgboost version used: python 0.6

xgboost.__version__
'0.6'

If you are using python package, please provide

1. The python version and distribution
2. The command to install xgboost if you are not installing from source
   ikamenshchikov@giskard-remote:~/MLBootCampV$ python3 -V
   Python 3.5.3
   pip3 install xgboost

Steps to reproduce

1. XGBClassifier(booster = 'gbtree', objective='binary:logistic')

TypeError: __init__() got an unexpected keyword argument 'booster'

What have you tried?

1. passing arguement as constructor
2. passing arguement with parameters (clf.train(params,..

Invalid parameter booster for estimator XGBClassifier. Check the list of available parameters w
ith `estimator.get_params().keys()

1. checked 'estimator.get_params().keys():

dict_keys(['min_child_weight', 'learning_rate', 'objective', 'subsample', 'reg_alpha', 'gamma', 'missing', 
'n_estimators', 'silent', 'reg_lambda', 'max_depth', 'colsample_bytree', 'max_delta_step', 'colsample_bylev
el', 'base_score', 'scale_pos_weight', 'seed', 'nthread'])

booster is not among them!

[yukw777]

I'm seeing this issue as well.

[morganmcg1]

Yes I am also having this issue

[khotilov]

The pypi package is too old (the python package version version is currently meaningless). Install the latest code from github.

[transposition]

I have built the latest xgboost of vesion 0.6a2, but still not support "booster"

[FrugoFruit90]

I have the same issue 😢
Interestingly, it works on Windows, but not on CentOS (where I try to do hyperparameter optimization)