DTreeViz 命令期间训练数据出现 ValueError

Posted 2023-03-12

【中文标题】DTreeViz 命令期间训练数据出现 ValueError

【英文标题】：ValueError with Training Data during DTreeViz Command

【发布时间】：2021-08-27 01:20:41

【问题描述】：

我创建了一个 DecisionTreeClassifier clf 来对数据进行建模，并尝试使用 dtreeviz 包来可视化树。

clf = DecisionTreeClassifier(max_depth=3)
clf.fit(X_train, y_train)

为了让 dtreeviz 函数可以消化数据，我对 X_train 和 y_train 进行了转换。

from dtreeviz.trees import dtreeviz
from sklearn import preprocessing
import graphviz

# Create integer representation of target column
label_encoder = preprocessing.LabelEncoder()
label_encoder.fit(y_train)
print ("Categorical classes:", label_encoder.classes_)
y_train_encoded = label_encoder.transform(y_train)

X_train_mod = X_train.to_numpy()
print("X_train type: ", type(X_train_mod))
print("Dimensions: ", np.ndim(X_train_mod))
print(X_train_mod.shape)

Categorical classes: ['Bad' 'Good']\
X_train type:  <class 'numpy.ndarray'>\
Dimensions:  2\
(700, 61)

但是，dtreeviz 命令失败并出现以下错误：

dtreeviz(clf, x_data=X_train_mod, y_data=y_train_encoded, target_name='Good/Bad', 
         feature_names=X_train.columns.to_list(), class_names=list(label_encoder.classes_))

ValueError                                Traceback (most recent call last)
<ipython-input-130-eec9abdfba5d> in <module>
     14 print(X_train_mod.shape)
     15 
---> 16 dtreeviz(clf, x_data=X_train_mod, y_data=y_train_encoded, target_name='Good/Bad', 
     17          feature_names=X_train.columns.to_list(), class_names=list(label_encoder.classes_))

~/opt/anaconda3/lib/python3.8/site-packages/dtreeviz/trees.py in dtreeviz(tree_model, x_data, y_data, feature_names, target_name, class_names, tree_index, precision, orientation, instance_orientation, show_root_edge_labels, show_node_labels, show_just_path, fancy, histtype, highlight_path, X, max_X_features_LR, max_X_features_TD, label_fontsize, ticks_fontsize, fontname, title, title_fontsize, colors, scale)
    795         if fancy:
    796             if shadow_tree.is_classifier():
--> 797                 class_split_viz(node, X_data, y_data,
    798                                 filename=f"tmp/nodenode.id_os.getpid().svg",
    799                                 precision=precision,

~/opt/anaconda3/lib/python3.8/site-packages/dtreeviz/trees.py in class_split_viz(node, X_train, y_train, colors, node_heights, filename, ticks_fontsize, label_fontsize, fontname, precision, histtype, X, highlight_node)
   1002 
   1003         bins = _get_bins(overall_feature_range, nbins)
-> 1004         hist, bins, barcontainers = ax.hist(X_hist,
   1005                                             color=X_colors,
   1006                                             align='mid',

~/opt/anaconda3/lib/python3.8/site-packages/matplotlib/__init__.py in inner(ax, data, *args, **kwargs)
   1599     def inner(ax, *args, data=None, **kwargs):
   1600         if data is None:
-> 1601             return func(ax, *map(sanitize_sequence, args), **kwargs)
   1602 
   1603         bound = new_sig.bind(ax, *args, **kwargs)

~/opt/anaconda3/lib/python3.8/site-packages/matplotlib/axes/_axes.py in hist(self, x, bins, range, density, weights, cumulative, bottom, histtype, align, orientation, rwidth, log, color, label, stacked, normed, **kwargs)
   6686         input_empty = np.size(x) == 0
   6687         # Massage 'x' for processing.
-> 6688         x = cbook._reshape_2D(x, 'x')
   6689         nx = len(x)  # number of datasets
   6690 

~/opt/anaconda3/lib/python3.8/site-packages/matplotlib/cbook/__init__.py in _reshape_2D(X, name)
   1428         return [np.reshape(x, -1) for x in X]
   1429     else:
-> 1430         raise ValueError(" must have 2 or fewer dimensions".format(name))
   1431 
   1432 

ValueError: x must have 2 or fewer dimensions

但是，根据 np.ndim() 函数的输出，看起来 X_train 的尺寸是正确的。我已经与 iris 示例 here 进行了比较，以验证参数的所有类型是否匹配。我现在不知道该怎么做。

【参考方案1】：

我遇到了同样的问题。确保您的分类器也接受了编码标签的训练，即使用

clf.fit(X_train, y_train_encoded)

而不是

clf.fit(X_train, y_train)