Mockingjay
v0.1.0

Mockingjay.Tree (Mockingjay v0.1.0)

A struct containing a convenient in-memory representation of a decision tree.

Each "node" in the tree is a Tree struct. A "decision" or "inner" node is any node whose :left and :right values are a valid Tree. A "leaf" or "outer" node is any node whose :left and :right values are nil. :left and :right must either both be nil or both be a Tree.

  • :id - The id of the node. This is a unique reference. This is generated automatically when using the from_map function.
  • :left - The left child of the node. This is nil for leaf nodes. This is a Tree for decision nodes.
  • :right - The right child of the node. This is nil for leaf nodes. This is a Tree for decision nodes.
  • :value - The value of the node:
    • For leaf nodes, this is the value as a number.
    • For non-leaf nodes, this is a map containing the following keys:
      • :feature - The feature used to split the data (if it is not a leaf).
      • :threshold - The threshold used to split the data (if it is not a leaf).

Link to this section Summary

Types

t()

A simple binary tree implementation.

Functions

bfs(root)

Returns a list of nodes in BFS order. For the uses in Mockingjay, BFS is tree-level order from right to left on each level. The nodes include their children nodes.

child?(tree, child_id)

Checks is the given child_id exists in the tree.

depth(tree)

Returns the depth of the tree. The root node is at depth 0.

depth_from_level(tree, current_depth)

Returns the depth of the tree starting from the given level.

dfs(root)

Returns tree nodes as a list in DFS order.

from_map(t)

Returns a Tree struct from a map. The map must have the appropriate required keys for the Tree struct. Any extra keys are ignored.

get_decision_nodes(tree)

Returns a list of the decision nodes in the tree in BFS order.

get_decision_values(tree)

Returns a list of the values of the decision nodes in the tree in BFS order.

get_leaf_nodes(tree)

Returns a list of the leaf nodes in the tree in BFS order.

get_leaf_values(tree)

Returns a list of the values of the leaf nodes in the tree in BFS order.

reduce_tree(root, acc, reducer)

Traverse the tree in BFS order, applying a reducer function to each node.

Link to this section Types

Link to this type

t()

@type t() :: %Mockingjay.Tree{
  id: reference(),
  left: t() | nil,
  right: t() | nil,
  value: number() | %{feature: pos_integer(), threshold: number()}
}

A simple binary tree implementation.

Link to this section Functions

Link to this function

bfs(root)

Returns a list of nodes in BFS order. For the uses in Mockingjay, BFS is tree-level order from right to left on each level. The nodes include their children nodes.

Link to this function

child?(tree, child_id)

Checks is the given child_id exists in the tree.

Link to this function

depth(tree)

Returns the depth of the tree. The root node is at depth 0.

Link to this function

depth_from_level(tree, current_depth)

Returns the depth of the tree starting from the given level.

Link to this function

dfs(root)

Returns tree nodes as a list in DFS order.

Link to this function

from_map(t)

Returns a Tree struct from a map. The map must have the appropriate required keys for the Tree struct. Any extra keys are ignored.

Link to this function

get_decision_nodes(tree)

Returns a list of the decision nodes in the tree in BFS order.

Link to this function

get_decision_values(tree)

Returns a list of the values of the decision nodes in the tree in BFS order.

Link to this function

get_leaf_nodes(tree)

Returns a list of the leaf nodes in the tree in BFS order.

Link to this function

get_leaf_values(tree)

Returns a list of the values of the leaf nodes in the tree in BFS order.

Link to this function

reduce_tree(root, acc, reducer)

Traverse the tree in BFS order, applying a reducer function to each node.