Tree classifier to graphviz ERROR - python-3.x

Question

I had made a Tree Classifier named model and tried to use the export graphviz function like this:
export_graphviz(decision_tree=model,
out_file='NT_model.dot',
feature_names=X_train.columns,
class_names=model.classes_,
leaves_parallel=True,
filled=True,
rotate=False,
rounded=True)
For some reason my run had raised this exception:
TypeError Traceback (most recent call last)
<ipython-input-298-40fe56bb0c85> in <module>()
6 filled=True,
7 rotate=False,
----> 8 rounded=True)
C:\Users\yonatanv\AppData\Local\Continuum\Anaconda3\lib\site-
packages\sklearn\tree\export.py in export_graphviz(decision_tree, out_file,
max_depth, feature_names, class_names, label, filled, leaves_parallel,
impurity, node_ids, proportion, rotate, rounded, special_characters)
431 recurse(decision_tree, 0, criterion="impurity")
432 else:
--> 433 recurse(decision_tree.tree_, 0,
criterion=decision_tree.criterion)
434
435 # If required, draw leaf nodes at same depth as each other
C:\Users\yonatanv\AppData\Local\Continuum\Anaconda3\lib\site-
packages\sklearn\tree\export.py in recurse(tree, node_id, criterion, parent,
depth)
319 out_file.write('%d [label=%s'
320 % (node_id,
--> 321 node_to_str(tree, node_id,
criterion)))
322
323 if filled:
C:\Users\yonatanv\AppData\Local\Continuum\Anaconda3\lib\site-
packages\sklearn\tree\export.py in node_to_str(tree, node_id, criterion)
289 np.argmax(value),
290 characters[2])
--> 291 node_string += class_name
292
293 # Clean up any trailing newlines
TypeError: ufunc 'add' did not contain a loop with signature matching types
dtype('<U90') dtype('<U90') dtype('<U90')
My hyper parameters for the visualizations are those:
print(model)
DecisionTreeClassifier(class_weight={1.0: 10, 0.0: 1}, criterion='gini',
max_depth=7, max_features=None, max_leaf_nodes=None,
min_impurity_split=1e-07, min_samples_leaf=50,
min_samples_split=2, min_weight_fraction_leaf=0.0,
presort=False, random_state=0, splitter='best')
print(model.classes_)
[ 0. , 1. ]
Help would be most appreciated!

Answer 1

As you see here specified in the documentation of export_graphviz, the param class_names works for strings, not float or int.
class_names : list of strings, bool or None, optional (default=None)
Try converting the model.classes_ to list of strings before passing them in export_graphviz.
Try class_names=['0', '1'] or class_names=['0.0', '1.0'] in the call to export_graphviz().
For a more general solution, use:
class_names=[str(x) for x in model.classes_]
But is there a specific reason that you are passing float values as y in model.fit()? Because that is mostly not required in classification task. Do you have actual y labels as this only or are you converting string labels to numeric before fitting the model?