Developer Guide and Reference

  • 2021.2
  • 03/26/2021
  • Public Content
Contents

Regression Decision Tree

Regression decision tree is a kind of decision trees described in Classification and Regression > Decision Tree.

Details

Given:
  • n feature vectors LaTex Math image. of size p
  • The vector of responses LaTex Math image. , where LaTex Math image. describes the dependent variable for independent variables LaTex Math image..
The problem is to build a regression decision tree.
Split Criterion
The library provides the decision tree regression algorithm based on the mean-squared error (MSE) [Breiman84]:
LaTex Math image.
Where
  • LaTex Math image. is the set of all possible outcomes of test LaTex Math image.
  • LaTex Math image. is the subset of
    D
    , for which outcome of LaTex Math image. is
    v
    , for example, LaTex Math image..
The test used in the node is selected as LaTex Math image.. For binary decision tree with “true” and “false” branches,
LaTex Math image.
Training Stage
The regression decision tree follows the algorithmic framework of decision tree training described in Decision Tree.
Prediction Stage
The regression decision tree follows the algorithmic framework of decision tree prediction described in Decision Tree.
Given the regression decision tree and vectors LaTex Math image., the problem is to calculate the responses for those vectors.

Batch Processing

Decision tree regression follows the general workflow described in Regression Usage Model.
Training
At the training stage, decision tree regression has the following parameters:
Parameter
Default Value
Description
algorithmFPType
float
The floating-point type that the algorithm uses for intermediate computations. Can be
float
or
double
.
method
defaultDense
The computation method used by the decision tree regression. The only training method supported so far is the default dense method.
pruning
reducedErrorPruning
Method to perform post-pruning. Available options for the pruning parameter:
  • reducedErrorPruning
    - reduced error pruning. Provide dataForPruning and dependentVariablesForPruning inputs, if you use pruning.
  • none
    - do not prune.
maxTreeDepth
0
Maximum tree depth. Zero value means unlimited depth. Can be any non-negative number.
minObservationsInLeafNodes
5
Minimum number of observations in the leaf node. Can be any positive number.
pruningFraction
0.2
Fraction of observations from training dataset to be used as observations for post-pruning via random sampling. The rest observations (with fraction LaTex Math image. to be used to build a decision tree). Can be any number in the interval (0, 1). If pruning is not used, all observations are used to build the decision tree regardless of this parameter value.
engine
SharedPtr<engines::mt19937::Batch<> >()
Pointer to the random number engine to be used for random sampling for reduced error post-pruning.
Prediction
At the prediction stage, decision tree regression has the following parameters:
Parameter
Default Value
Description
algorithmFPType
float
The floating-point type that the algorithm uses for intermediate computations. Can be
float
or
double
.
method
defaultDense
The computation method used by the decision tree regression. The only training method supported so far is the default dense method.
Examples
C++ (CPU)
Batch Processing:
Java*
There is no support for Java on GPU.
Batch Processing:

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.