dtreeviz

dtreeviz is a Python library for decision tree visualization and model interpretation. It supports several machine learning libraries including scikit-learn, XGBoost, Spark MLlib, LightGBM, and TensorFlow (Decision Forests).

To use dtreeviz, you follow a standard workflow:

1. Import dtreeviz and your ML library.
2. Load your data.
3. Train your model.
4. Create a dtreeviz adaptor using dtreeviz.model().
5. Use the adaptor to call visualization methods like .view() or .explain_prediction_path().

from sklearn.datasets import load_iris
from sklearn.tree import DecisionTreeClassifier
import dtreeviz

iris = load_iris()
X = iris.data
y = iris.target

clf = DecisionTreeClassifier(max_depth=4)
clf.fit(X, y)

viz_model = dtreeviz.model(clf,
                           X_train=X, y_train=y,
                           feature_names=iris.feature_names,
                           target_name='iris',
                           class_names=iris.target_names)

v = viz_model.view()     # render as SVG into internal object 
v.show()                 # pop up window
v.save("/tmp/iris.svg")  # optionally save as svg

Install dtreeviz using pip. You can install specific extras depending on the machine learning library you are using. It is recommended to use an Anaconda Prompt on Windows.

Note: Ensure you do not have conda-installed graphviz packages (like python-graphviz or graphviz) as dtreeviz requires the pip versions. You can remove them using:

conda uninstall python-graphviz
conda uninstall graphviz

pip install dtreeviz             # install dtreeviz for sklearn
pip install dtreeviz[xgboost]    # install XGBoost related dependency
pip install dtreeviz[pyspark]    # install pyspark related dependency
pip install dtreeviz[lightgbm]   # install LightGBM related dependency
pip install dtreeviz[tensorflow_decision_forests]   # install tensorflow_decision_forests related dependency
pip install dtreeviz[ai]         # install AI chat/explanation features (requires OpenAI API key)
pip install dtreeviz[all]        # install all related dependencies

If you enable AI integration, you can interact with your decision tree using the .chat() method. The AI can answer questions regarding:

• Tree structure: Architecture, depth, node count, and splitting criteria.
• Tree nodes: Split conditions, feature usage, and node statistics.
• Leaf nodes: Predictions, confidence scores, and class distributions.
• Training dataset: Feature statistics and data characteristics within nodes.

When ai_chat=True is enabled, the standard .view() method will also automatically include an LLM-generated explanation alongside the visual output.

Requirements:

• Install with pip install dtreeviz[ai].
• Set the OPENAI_API_KEY environment variable.

# Enable AI chat when creating the model
viz_model = dtreeviz.model(tree_classifier,
                           X_train=dataset[features], y_train=dataset[target],
                           feature_names=features,
                           target_name=target, class_names=["perish", "survive"],
                           ai_chat=True,
                           ai_model="gpt-4.1-mini",
                           max_history_messages=10)

# Ask questions about your tree
viz_model.chat("Please give me a short summary of the tree structure?")
viz_model.chat("Which leaf nodes have the lowest prediction confidence?")