importargparseimportnumpyasnpfrommatplotlibimportpyplotaspltimportxgboostdefmain(args:argparse.Namespace)->None:rng=np.random.RandomState(1994)kRows=4196kCols=10X=rng.randn(kRows,kCols)y=rng.randn(kRows)fw=np.ones(shape=(kCols,))foriinrange(kCols):fw[i]*=float(i)dtrain=xgboost.DMatrix(X,y)dtrain.set_info(feature_weights=fw)# Perform column sampling for each node split evaluation, the sampling process is# weighted by feature weights.bst=xgboost.train({"tree_method":"hist","colsample_bynode":0.2},dtrain,num_boost_round=10,evals=[(dtrain,"d")],)feature_map=bst.get_fscore()# feature zero has 0 weightassertfeature_map.get("f0",None)isNoneassertmax(feature_map.values())==feature_map.get("f9")ifargs.plot:xgboost.plot_importance(bst)plt.show()if__name__=="__main__":parser=argparse.ArgumentParser()parser.add_argument("--plot",type=int,default=1,help="Set to 0 to disable plotting the evaluation history.",)args=parser.parse_args()main(args)

DownloadJupyternotebook:feature_weights.ipynb

DownloadPythonsourcecode:feature_weights.py

Downloadzipped:feature_weights.zip