import numpy as np
from sklearn.pipeline import Pipeline
from sklearn.svm import SVC
from sklearn.ensemble import BaggingClassifier
#------------------------------------------------------------------------------
def main():
    #1) prepare data
    np.random.seed(0)
    feat=np.random.normal(size=(1000,2))
    lbl=np.random.choice(a=2,p=[.5,.5],size=1000,replace=True)
    wghts=np.ones(shape=(1000,))
    
    #2) fit using standard SVC
    bg0=BaggingClassifier(base_estimator=SVC(probability=True,random_state=0),n_estimators=100,max_samples=1.,n_jobs=-1,random_state=0)
    bg0=bg0.fit(feat,lbl)
    prob0=bg0.predict_proba(feat)
    #6.1) fit using standard pipe

    #3) fit using a Pipeline that contains only a SVC
    bg1=BaggingClassifier(base_estimator=Pipeline([('clf',SVC(probability=True,random_state=0))]),n_estimators=100,max_samples=1.,n_jobs=-1,random_state=0)
    bg1=bg1.fit(feat,lbl)
    prob1=bg1.predict_proba(feat)
    print 'match?',np.allclose(prob0,prob1)
#------------------------------------------------------------------------------
if __name__=='__main__':main()

bg0=bg0.fit(feat,lbl,wghts) # works!
bg1=bg1.fit(feat,lbl,wghts) # breaks...
ValueError: The base estimator doesn't support sample weight