dtreeviz: Decision Tree Visualization

2.3.2 verified Fri May 01 auth: no python

A Python 3 library for visualizing decision trees from scikit-learn, XGBoost, LightGBM, Spark, and TensorFlow. Version 2.3.2 supports AI chat integration for sklearn, categorical variables, and various tree model backends. Releases are frequent, approximately every few months.

pip install dtreeviz

Common errors

error TypeError: dict() argument after ** must be a mapping, not float ↓

cause Incompatibility with newer versions of numpy or other dependencies, fixed in 2.2.1.

fix

Upgrade dtreeviz to >=2.2.1 or pin numpy to a compatible version.

error KeyError when using decision_boundaries function ↓

cause Bug in version 2.0.x, fixed in 2.1.0.

fix

Upgrade dtreeviz to >=2.1.0.

Warnings

breaking Version 2.1.0 introduced a major refactoring. Functions like 'ctree_feature_space' changed signature; older code may break. ↓

fix Update function calls to match new signatures. Refer to changelog for specific changes.

gotcha When using categorical features, ensure they are numeric-encoded. dtreeviz does not handle string categories natively in older versions; supported from 2.2.0 onward. ↓

fix Upgrade to >=2.2.0 or encode categories as integers before fitting.

Imports

dtreeviz
wrong
```
import dtreeviz
```
correct
```
from dtreeviz import dtreeviz
```
The main function is inside the dtreeviz module. Using 'import dtreeviz' then calling dtreeviz.dtreeviz(...) is also valid but less common.

model

from sklearn.tree import DecisionTreeRegressor

Quickstart

Train a simple decision tree regressor and visualize it with dtreeviz.

from sklearn.datasets import load_diabetes
from sklearn.tree import DecisionTreeRegressor
from dtreeviz import dtreeviz

diabetes = load_diabetes()
X = diabetes.data
y = diabetes.target

regr = DecisionTreeRegressor(max_depth=3)
regr.fit(X, y)

viz = dtreeviz(regr, X, y, target_name='diabetes', feature_names=diabetes.feature_names)
viz.view()