Python中Graphviz使用问题：程序终止

Hey there! Let's break down why your visualization code is crashing and get it working properly. From what you shared, there are a few common issues that could be causing the problem—let's go through them one by one.

Common Issues & Fixes #

1. Incomplete Method Call #

Your code cuts off at graph = pydotplus.graph_from_dot_d...—this is a syntax error right off the bat. The full, correct method call needs to fetch the DOT data from your StringIO object:

graph = pydotplus.graph_from_dot_data(dot_data.getvalue())

2. Deprecated sklearn.externals.six Module #

In newer versions of scikit-learn (0.23+), sklearn.externals.six has been removed. You should replace that import with Python's standard library StringIO:

# Replace this:
# from sklearn.externals.six import StringIO
# With this:
from io import StringIO

3. Missing Graphviz System Dependency #

pydotplus acts as a Python wrapper for Graphviz, but it doesn't include the actual Graphviz binary. If you haven't installed Graphviz on your system, the code will fail to generate the image. Here's how to fix it:

• Windows: Download the Graphviz installer, run it, and make sure to add Graphviz's bin folder to your system PATH.
• macOS: Use Homebrew: brew install graphviz
• Linux (Debian/Ubuntu): Use apt: sudo apt-get install graphviz

Also, double-check you have the required Python packages installed:

pip install pydotplus graphviz

Full Working Code Example #

Here's the corrected, complete code that should run without crashing:

from sklearn import tree
from io import StringIO
import pydotplus
from IPython.display import Image

# Assume your model is already trained with X and Y
model = tree.DecisionTreeClassifier()
model = model.fit(X, Y)

# Generate DOT data
dot_data = StringIO()
tree.export_graphviz(
    model,
    out_file=dot_data,
    filled=True,
    rounded=True,
    special_characters=True,
    leaves_parallel=False
)

# Create graph from DOT data
graph = pydotplus.graph_from_dot_data(dot_data.getvalue())

# Save as PNG file
graph.write_png("decision_tree.png")

# Optional: Display the image in IPython/Jupyter
Image(graph.create_png())

If you still run into issues, check the exact error message—usually it'll tell you if it's a missing dependency or syntax problem. For example, a "dot: command not found" error means Graphviz isn't installed on your system.

内容的提问来源于stack exchange，提问作者F.Ferlande