sklearn.externals.joblib无法加载pkl文件

Question

I inherited some code from a coworker who left my company, and I noticed that an email open predictor he wrote hasn't run in a few weeks. So, I tried to run it. It got to this line:

clf_predict = joblib.load(os.path.join(data_dir,'trained_random_forest.pkl'))

And the result was this:

File "/usr/local/lib/python2.7/distpackages/sklearn/externals/joblib/numpy_pickle.py", line 425, in load
obj = unpickler.load()
File "/usr/lib/python2.7/pickle.py", line 858, in load dispatch[key](self)
File "/usr/lib/python2.7/pickle.py", line 1133, in load_reduce
value = func(*args)
File "sklearn/tree/_tree.pyx", line 923, in sklearn.tree._tree.Splitter.__cinit__ (sklearn/tree/_tree.c:8329)
TypeError: __cinit__() takes exactly 5 positional arguments (4 given)

The syntax for the load looks right to me from what I've seen in the documentation. Therefore, my initial theory is that maybe scikit-learn was upgraded and can't read the current pickle file (which is a few months old at this point). However, I don't know for sure that it was upgraded, and I'm pretty new to working with sklearn, so I don't know if that would pose a problem. Appreciate any pointers you can give

Answer 1

Assuming that it worked for your co-worker and your problem is due to an sklearn upgrade. The extra argument to the Splitter init was introduced in this commit . You could try loading the pkl file in an earlier version of sklearn and take it from there.