2.2.2

See https://github.com/parrt/dtreeviz/milestone/40?closed=1

Contributors

• 2 items: Tudor Lapusan (tlapusan)
• 1 items: StepanWorkV (StepanWorkV)
• 1 items: Tyrion (hxhxd)

2.2.1

Issues fixed

• TypeError: dict() argument after ** must be a mapping, not float (compatibility)
• Visualisation error when a node will contains samples from only one class (sklearn) (bug)
• Pie charts show for data that classifier was trained on and not new data (enhancement)

Improvements, features

• Make sklearn visualisations to support validation datasets. (enhancement)
• Pie charts show for data that classifier was trained on and not new data (enhancement)
• Add different display types for leaf_distributions x-axis. (enhancement)

Contributors

• 4 items: Tudor Lapusan (tlapusan)
• 1 items: Alex Moldovan (Alex-AMC)
• 1 items: Rajdeep (RajdeepTarafder)
• 1 items: HannahAlexander
• 1 items: Matthew Epland (mepland)

2.2.0

Support string categorical vars in regressors and classifiers. See #265

2.1.4

Minor improvement release: #259

2.1.3

Add doc for feature space partitioning functions. Fix bug in tessellate() function.

2.1.2

Fix compatibility issue with tensorflow trees.

2.1.1

Single feature release:

Enhanced ctree_feature_space() that accepts a features arg rather than requiring a model with just 1 or 2 features

2.1.0

This is a major cleanup release, with @mepland doing most of the heavy lifting; there are refactorings, if you bug fixes, and lots of visualization clean up.

Issues fixed

• TF handle multi class labels for view visualisation (bug)
• Standardize Capitalization (clean up)
• KeyError when using decision_boundaries function (clean up)
• Dimension issue in dtreeviz_sklearn_pipeline_visualisations.ipynb (compatibility)
• font name/size not respected (bug)
• Boston dataset no longer avaliable in sklearn for classifier-decision-boundaries.ipynb and classifier-boundary-animations.ipynb (clean up)
• ctree_leaf_distributions() should use the classnames not classes (bug)
• rtreeviz_univar() does not show X feature properly (bug)
• class name font could not change (bug)
• labels too close to the axes overlap for classifiers (enhancement)

Improvements, features

• Order explain_prediction_path() (enhancement)
• Improve _ctreeviz_univar() (enhancement)
• Fix Digraph Graphics (enhancement)
• Allow None legend edge (enhancement)
• ctree_leaf_distributions() label_all_leaves parameter (enhancement)
• Allow different left and right arrow colors via 'larrow' and 'rarrow' (enhancement)
• Add matplotlib default colors as mpl_colors (enhancement)
• Add option to plot classifier leaves as horizontal bars (enhancement)
• labels too close to the axes overlap for classifiers (enhancement)

Contributors

• 19 items: Matthew Epland (mepland)
• 5 items: Terence Parr (parrt)
• 2 items: Tudor Lapusan (tlapusan)
• 1 items: 0ptimista
• 1 items: Tobias Windisch (windisch)
• 1 items: HidenariUoi

2.0.0

This release re-organizes the API to focus on using a model adaptor that adapts the visualization library to the various supported decision tree libraries.

We simplified the README and rebuilt all of the library-specific notebooks to demonstrate the new API, using a common set of examples:

• sklearn-based examples (colab)
• LightGBM-based examples (colab)
• Spark-based examples (colab)
• TensorFlow-based examples (colab)
• XGBoost-based examples (colab)
• Classifier decision boundaries for any scikit-learn model.ipynb (colab)
• Changing colors notebook (specific to sklearn) (colab)

New API:

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

Previous API:
Previously, we did something like this to call functions and pass in the various details of the model and training data:

from dtreeviz.trees import dtreeviz
dtreeviz(tree_model=clf, X_train, ...)

Using old functions with 2.0+:

For backward compatibility to call function dtreeviz() and the old API, you can change the import to be:

from dtreeviz import *
dtreeviz(tree_model=clf, X_train, ...)

Argument name changes:

If you were previously using internal model adaptors, such as ShadowLightGBMTree, please note we have changed the following argument names: x_data->X_train and y_data->y_train.

Stuff we completed:

https://github.com/parrt/dtreeviz/milestone/30?closed=1

Fix bar chart viz weirdness

Fixing this bug: #196