如何在 Windows 上的 python 中安装 XGBoost 包

Question

I tried to install XGBoost package in python. I am using windows os, 64bits . I have gone through following.

The package directory states that xgboost is unstable for windows and is disabled: pip installation on windows is currently disabled for further invesigation, please install from github. https://pypi.python.org/pypi/xgboost/

I am not well versed in Visual Studio, facing problem building XGBoost. I am missing opportunities to utilize xgboost package in data science.

Please guide, so that I can import the XGBoost package in python.

Thanks

Answer 1

If you are using anaconda (or miniconda ) you can use the following:

• conda install -c anaconda py-xgboost UPDATED 2019-09-20
• Docs

Check install by:

• Activating the environment (see below)
• Running conda list

---

To activate an environment :

On Windows, in your Anaconda Prompt, run (assumes your environment is named myenv ):

• activate myenv

On macOS and Linux, in your Terminal Window, run (assumes your environment is named myenv ):

• source activate myenv

Conda prepends the path name myenv onto your system command.

Answer 2

Build it from here:

• download xgboost whl file from here (make sure to match your python version and system architecture, eg "xgboost-0.6-cp35-cp35m-win_amd64.whl" for python 3.5 on 64-bit machine)
• open command prompt
• cd to your Downloads folder (or wherever you saved the whl file) pip install xgboost-0.6-cp35-cp35m-win_amd64.whl (or whatever your whl file is named)

Answer 3

You first need to build the library through "make", then you can install using anaconda prompt (if you want it on anaconda) or git bash (if you use it in Python only).

First follow the official guide with the following procedure (in Git Bash on Windows):

git clone --recursive https://github.com/dmlc/xgboost
git submodule init
git submodule update

then install TDM-GCC here and do the following in Git Bash:

alias make='mingw32-make'
cp make/mingw64.mk config.mk; make -j4

Last, do the following using anaconda prompt or Git Bash:

cd xgboost\python-package  
python setup.py install 

Also refer to these great resources:

Official Guide

Installing Xgboost on Windows

Installing XGBoost For Anaconda on Windows

Answer 4

You can pip install catboost. It is a recently open-sourced gradient boosting library, which is in most cases more accurate and faster than XGBoost, and it has categorical features support. Here is the site of the library: https://catboost.ai

Answer 5

The following command should work but, If you have a problem with this command

conda install -c conda-forge xgboost

First activate your environment . Assume your environment is named simply write in conda terminal :

activate <MY_ENV>

and then

 pip install xgboost

Answer 6

在 macOS 上，以下命令可以运行 conda install -c conda-forge xgboost 但在此之前我已经阅读了一些其他文章，因此确实使用 brew 安装了 gcc

Answer 7

pip install xgboost也适用于python 3.8 ，而上面提到的其他选项对我不起作用

Answer 8

I have installed xgboost in windows os following the above resources, which is not available till now in pip. However, I tried with the following function code, to get cv parameters tuned:

#Import libraries:
import pandas as pd
import numpy as np
import xgboost as xgb
from xgboost.sklearn import XGBClassifier
from sklearn import cross_validation, metrics   #Additional sklearn functions
from sklearn.grid_search import GridSearchCV   #Perforing grid search

import matplotlib.pylab as plt
%matplotlib inline
from matplotlib.pylab import rcParams
rcParams['figure.figsize'] = 12, 4

train = pd.read_csv('train_data.csv')
target = 'target_value'
IDcol = 'ID'

A function is created to get the optimum parameters and display the output in visual form.

def modelfit(alg, dtrain, predictors,useTrainCV=True, cv_folds=5, early_stopping_rounds=50):

if useTrainCV:
    xgb_param = alg.get_xgb_params()
    xgtrain = xgb.DMatrix(dtrain[predictors].values, label=dtrain[target].values)
    cvresult = xgb.cv(xgb_param, xgtrain, num_boost_round=alg.get_params()['n_estimators'], nfold=cv_folds,
        metrics='auc', early_stopping_rounds=early_stopping_rounds, show_progress=False)
    alg.set_params(n_estimators=cvresult.shape[0])

#Fit the algorithm on the data
alg.fit(dtrain[predictors], dtrain[target_label],eval_metric='auc')

#Predict training set:
dtrain_predictions = alg.predict(dtrain[predictors])
dtrain_predprob = alg.predict_proba(dtrain[predictors])[:,1]

#Print model report:
print "\nModel Report"
print "Accuracy : %.4g" % metrics.accuracy_score(dtrain[target_label].values, dtrain_predictions)
print "AUC Score (Train): %f" % metrics.roc_auc_score(dtrain[target_label], dtrain_predprob)

feat_imp = pd.Series(alg.booster().get_fscore()).sort_values(ascending=False)
feat_imp.plot(kind='bar', title='Feature Importances')
plt.ylabel('Feature Importance Score')

Now, when the function is called to get the optimum parameters:

  #Choose all predictors except target & IDcols
  predictors = [x for x in train.columns if x not in [target]]
  xgb = XGBClassifier(
  learning_rate =0.1,
  n_estimators=1000,
  max_depth=5,
  min_child_weight=1,
  gamma=0,
  subsample=0.7,
  colsample_bytree=0.7,
  objective= 'binary:logistic',
  nthread=4,
  scale_pos_weight=1,
  seed=198)
 modelfit(xgb, train, predictors)

Although the feature importance chart is displayed, but the parameters info in red box at the top of chart is missing: Consulted people who use linux/mac OS and got xgboost installed. They are getting the above info. I was wondering whether it is due to specific implementation , I build and installed in windows. And how I can get the parameters info displayed above the chart. As of now, I am getting the chart and not the red box and info within it. Thanks.

Answer 9

Besides what's already on developers' github, which is building from source(creating a c++ environment, etc.), I have found an easier way to do it, which I explained here with details. Basically, you have to go a website by UC Irvine and download a .whl file, then cd to the folder and install xgboost with pip.

Answer 10

XGBoost is used in Applied Machine Learning and is known for its gradient boost algorithm and it is available as a library in python but has to be compiled using cmake .

Alternatively what you can do is from this link you can download the C pre-compiled library and install it using the pip install < FILE-NAME.whl> command. Ensure you have downloaded the library which is compatible with your python version.

I experienced this problem while I was using the same in Anaconda(Spyder). Then just restart the kernel and your error will go away.