在XGBoost版本0.6（Python）中访问功能重要性

Question

I'm trying to to get the feature importance for my xgboost model but I'm always getting this error:

AttributeError: 'XGBModel' object has no attribute 'feature_importances_'

I read that this might be a versioning issue, but I installed this in late December 2016 and I believe this is the version that supposedly fixed this problem.

Can anyone help me get the feature importances?

import numpy as np
import pandas as pd
import xgboost as xgb
print "XGB Version: " + str(xgb.__version__)

np.random.seed(0)
df = pd.DataFrame(np.random.randint(0,1000,size=(1000, 5)), columns=list('ABCDE'))

x = df.ix[:,0:4]
y = df.ix[:,4]

mod = xgb.XGBModel()
mod.fit (x,y)

mod.feature_importances_

Answer 1

Nothing wrong, it work perfect