Skip to content


Hoeffding Tree or Very Fast Decision Tree classifier.


  • grace_period

    Type → int

    Default → 200

    Number of instances a leaf should observe between split attempts.

  • max_depth

    Type → int | None

    Default → None

    The maximum depth a tree can reach. If None, the tree will grow indefinitely.

  • split_criterion

    Type → str

    Default → info_gain

    Split criterion to use.
    - 'gini' - Gini
    - 'info_gain' - Information Gain
    - 'hellinger' - Helinger Distance

  • delta

    Type → float

    Default → 1e-07

    Significance level to calculate the Hoeffding bound. The significance level is given by 1 - delta. Values closer to zero imply longer split decision delays.

  • tau

    Type → float

    Default → 0.05

    Threshold below which a split will be forced to break ties.

  • leaf_prediction

    Type → str

    Default → nba

    Prediction mechanism used at leafs.
    - 'mc' - Majority Class
    - 'nb' - Naive Bayes
    - 'nba' - Naive Bayes Adaptive

  • nb_threshold

    Type → int

    Default → 0

    Number of instances a leaf should observe before allowing Naive Bayes.

  • nominal_attributes

    Type → list | None

    Default → None

    List of Nominal attributes identifiers. If empty, then assume that all numeric attributes should be treated as continuous.

  • splitter

    Type → Splitter | None

    Default → None

    The Splitter or Attribute Observer (AO) used to monitor the class statistics of numeric features and perform splits. Splitters are available in the tree.splitter module. Different splitters are available for classification and regression tasks. Classification and regression splitters can be distinguished by their property is_target_class. This is an advanced option. Special care must be taken when choosing different splitters. By default, tree.splitter.GaussianSplitter is used if splitter is None.

  • binary_split

    Type → bool

    Default → False

    If True, only allow binary splits.

  • min_branch_fraction

    Type → float

    Default → 0.01

    The minimum percentage of observed data required for branches resulting from split candidates. To validate a split candidate, at least two resulting branches must have a percentage of samples greater than min_branch_fraction. This criterion prevents unnecessary splits when the majority of instances are concentrated in a single branch.

  • max_share_to_split

    Type → float

    Default → 0.99

    Only perform a split in a leaf if the proportion of elements in the majority class is smaller than this parameter value. This parameter avoids performing splits when most of the data belongs to a single class.

  • max_size

    Type → float

    Default → 100.0

    The max size of the tree, in Megabytes (MB).

  • memory_estimate_period

    Type → int

    Default → 1000000

    Interval (number of processed instances) between memory consumption checks.

  • stop_mem_management

    Type → bool

    Default → False

    If True, stop growing as soon as memory limit is hit.

  • remove_poor_attrs

    Type → bool

    Default → False

    If True, disable poor attributes to reduce memory usage.

  • merit_preprune

    Type → bool

    Default → True

    If True, enable merit-based tree pre-pruning.


  • height

  • leaf_prediction

    Return the prediction strategy used by the tree at its leaves.

  • max_size

    Max allowed size tree can reach (in MB).

  • n_active_leaves

  • n_branches

  • n_inactive_leaves

  • n_leaves

  • n_nodes

  • split_criterion

    Return a string with the name of the split criterion being used by the tree.

  • summary

    Collect metrics corresponding to the current status of the tree in a string buffer.


from river.datasets import synth
from river import evaluate
from river import metrics
from river import tree

gen = synth.Agrawal(classification_function=0, seed=42)
dataset = iter(gen.take(1000))

model = tree.HoeffdingTreeClassifier(
    nominal_attributes=['elevel', 'car', 'zipcode']

metric = metrics.Accuracy()

evaluate.progressive_val_score(dataset, model, metric)
Accuracy: 84.58%



Print an explanation of how x is predicted.


  • x — 'dict'


str | None: A representation of the path followed by the tree to predict x; None if


Draw the tree using the graphviz library.

Since the tree is drawn without passing incoming samples, classification trees will show the majority class in their leaves, whereas regression trees will use the target mean.


  • max_depth — 'int | None' — defaults to None
    The maximum depth a tree can reach. If None, the tree will grow indefinitely.


Train the model on instance x and corresponding target y.


  • x
  • y
  • w — defaults to 1.0


Predict the label of a set of features x.


  • x — 'dict'
  • kwargs


base.typing.ClfTarget | None: The predicted label.


Predict the probability of each label for a dictionary of features x.


  • x


A dictionary that associates a probability which each label.


Return a representation of the current tree structure organized in a pandas.DataFrame object.

In case the tree is empty or it only contains a single node (a leaf), None is returned.




A Hoeffding Tree 1 is an incremental, anytime decision tree induction algorithm that is capable of learning from massive data streams, assuming that the distribution generating examples does not change over time. Hoeffding trees exploit the fact that a small sample can often be enough to choose an optimal splitting attribute. This idea is supported mathematically by the Hoeffding bound, which quantifies the number of observations (in our case, examples) needed to estimate some statistics within a prescribed precision (in our case, the goodness of an attribute).

A theoretically appealing feature of Hoeffding Trees not shared by other incremental decision tree learners is that it has sound guarantees of performance. Using the Hoeffding bound one can show that its output is asymptotically nearly identical to that of a non-incremental learner using infinitely many examples. Implementation based on MOA 2.

  1. G. Hulten, L. Spencer, and P. Domingos. Mining time-changing data streams. In KDD’01, pages 97–106, San Francisco, CA, 2001. ACM Press. 

  2. Albert Bifet, Geoff Holmes, Richard Kirkby, Bernhard Pfahringer. MOA: Massive Online Analysis; Journal of Machine Learning Research 11: 1601-1604, 2010.