Reduced Error Pruner

Module containing ReducedErrorPruner class.

class pruneabletree.pruner_rep.ReducedErrorPruner(tree)[source]

Pruner for decision trees that uses the Reduced Error Pruning (REP) technique [1] [2].

Note that the given tree is modified in place. To keep a copy of the original, clone it first.

Parameters:
tree : Tree object

The underlying tree object of a DecisionTreeClassifier (e.g. clf.tree_).

See also

pruneabletree.prune.PruneableDecisionTreeClassifier, pruneabletree.pruner_ebp.ErrorBasedPruner

References

[1](1, 2) J. Ross Quinlan. Simplifying decision trees. International journal of man-machine studies, 27(3):221-234, 1987.
[2](1, 2) Tapio Elomaa and Matti Kaariainen. An analysis of reduced error pruning. Journal of Artificial Intelligence Research, 15:163-187, 2001.

Methods

is_leaf(node_id) Returns True if the given node is a leaf.
leaf_prediction(node_id) Returns the class index of that the node with the given ID would predict.
num_actual_nodes(tree) Returns the actual number of nodes in the given tree after pruning.
num_instances(node_id[, y_idx]) Returns the number of instances in a given node.
num_leaves(tree) Returns the number of leaves.
prune(X_val, y_val) Prunes the given tree using the given validation set.
to_leaf(node_id, depths) Convert the node with the given ID to a leaf, pruning away its children.
prune(X_val, y_val)[source]

Prunes the given tree using the given validation set.

Parameters:
X_val : array-like or sparse matrix, shape = [n_samples, n_features]

The validation input samples. Internally, it will be converted to dtype=np.float32 and if a sparse matrix is provided to a sparse csc_matrix.

y_val : array-like, shape = [n_samples] or [n_samples, n_outputs]

The target values (class labels) as integers or strings.