How to Extract Decision Rules from scikit-learn Decision Trees?

Decision Rule Extraction from scikit-learn Decision Trees

Problem Statement:

Can the decision rules underlying a trained decision tree model be extracted as a textual list?

Solution:

Using the tree_to_code function, it is possible to generate a valid Python function that represents the decision rules of a scikit-learn decision tree:

<code class="python">from sklearn.tree import _tree

def tree_to_code(tree, feature_names):
    tree_ = tree.tree_
    feature_name = [
        feature_names[i] if i != _tree.TREE_UNDEFINED else "undefined!"
        for i in tree_.feature
    ]
    print("def tree({}):".format(", ".join(feature_names)))

    def recurse(node, depth):
        indent = "  " * depth
        if tree_.feature[node] != _tree.TREE_UNDEFINED:
            name = feature_name[node]
            threshold = tree_.threshold[node]
            print("{}if {} <= {}:".format(indent, name, threshold))
            recurse(tree_.children_left[node], depth + 1)
            print("{}else:  # if {} > {}".format(indent, name, threshold))
            recurse(tree_.children_right[node], depth + 1)
        else:
            print("{}return {}".format(indent, tree_.value[node]))

    recurse(0, 1)</code>

Example:

For a decision tree that attempts to return its input (a number between 0 and 10), the tree_to_code function would print the following Python function:

<code class="python">def tree(f0):
  if f0 <= 6.0:
    if f0 <= 1.5:
      return [[ 0.]]
    else:  # if f0 > 1.5
      if f0 <= 4.5:
        if f0 <= 3.5:
          return [[ 3.]]
        else:  # if f0 > 3.5
          return [[ 4.]]
      else:  # if f0 > 4.5
        return [[ 5.]]
  else:  # if f0 > 6.0
    if f0 <= 8.5:
      if f0 <= 7.5:
        return [[ 7.]]
      else:  # if f0 > 7.5
        return [[ 8.]]
    else:  # if f0 > 8.5
      return [[ 9.]]</code>

Caveats:

Avoid the following common issues:

• Do not rely on tree_.threshold == -2 to identify leaf nodes; check tree.feature or tree.children_* instead.
• Specify feature names correctly, avoiding integers that may correspond to undefined features.
• Single if statements suffice in the recursive function.

The above is the detailed content of How to Extract Decision Rules from scikit-learn Decision Trees?. For more information, please follow other related articles on the PHP Chinese website!