【Python】Scikit-learnのjoblib.dumpでエクスポートしたファイルをインポートできない

Question

問題

Scikit-learnのjoblib.dumpでエクスポートしたファイルをインポートすると以下のエラーが出ます。
解決策を教えて下さい。

Traceback (most recent call last):
  File "c:/python/test.py", line 26, in <module>
    clf2 = joblib.load(SAVE_FILE_PATH)
  File "C:\Users\xxx\AppData\Local\Programs\Python\Python38-32\lib\site-packages\joblib\numpy_pickle.py", line 605, in load
    obj = _unpickle(fobj, filename, mmap_mode)
  File "C:\Users\xxx\AppData\Local\Programs\Python\Python38-32\lib\site-packages\joblib\numpy_pickle.py", line 529, in _unpickle
    obj = unpickler.load()
  File "C:\Users\xxx\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1210, in load
    dispatch[key[0]](self)
  File "C:\Users\xxx\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1471, in load_dict
    items = self.pop_mark()
  File "C:\Users\xxx\AppData\Local\Programs\Python\Python38-32\lib\pickle.py", line 1217, in pop_mark
    self.stack = self.metastack.pop()
IndexError: pop from empty list

ソースコード、ファイル

import pandas as pd
from sklearn import tree
import joblib

LOAD_FILE_PATH = "C:/python/sample/input.csv"
SAVE_FILE_PATH = "C:/python/sample/output.clf"

data = pd.read_csv(LOAD_FILE_PATH, sep=",")

variables = ['x1', 'x2']

clf = tree.DecisionTreeClassifier()

clf = clf.fit(data[variables], data['x3'])

with open(SAVE_FILE_PATH, 'w') as f:
    f = tree.export_graphviz(clf, out_file=f)

clf2 = joblib.load(SAVE_FILE_PATH)

pred2 = clf2.predict(data[['x1', 'x2']])

print(pred2)

input.csv

x1,x2,x3
45,17.5,30
38,17.0,25
41,18.5,20
34,18.5,30
59,16.0,45
47,19.0,35
35,19.5,25
43,16.0,35
54,18.0,35
52,19.0,40

Answer 1

tree.export_graphvizではなくjoblib.dumpしたものを読み込みましょう。