爲DecisionTreeClassifier傳遞參數時出錯

Question

我嘗試使用參數在字符串中的DecisionTreeClassifier。

print d # d= 'max_depth=100' 
clf = DecisionTreeClassifier(d) 
clf.fit(X[:3000,], labels[:3000]) 

我在這種情況下得到低於錯誤。如果我使用clf = DecisionTreeClassifier(max_depth=100)它工作正常。

Traceback (most recent call last): 
    File "train.py", line 120, in <module> 
    grid_search_generalized(X, labels, {"max_depth":[i for i in range(100, 200)]}) 
    File "train.py", line 51, in grid_search_generalized 
    clf.fit(X[:3000,], labels[:3000]) 
    File "/usr/local/lib/python2.7/dist-packages/sklearn/tree/tree.py", line 790, in fit 
    X_idx_sorted=X_idx_sorted) 
    File "/usr/local/lib/python2.7/dist-packages/sklearn/tree/tree.py", line 326, in fit 
    criterion = CRITERIA_CLF[self.criterion](self.n_outputs_, 
KeyError: 'max_depth=100' 

Answer 1

您將參數作爲字符串對象傳遞，而不是作爲可選參數傳遞。
如果你真的有調用這個字符串的構造函數，你可以使用此代碼：

arg = dict([d.split("=")]) 
clf = DecisionTreeClassifier(**arg) 

你可以閱讀更多有關參數在這個環節
Passing a dictionary to a function in python as keyword parameters