🔗 RandomForest
The RandomForest
class implements a parallelized random forest classifier that
supports numerical and categorical features, by default using Gini gain to
choose which feature to split on in each tree.
Random forests are a collection of decision trees that give better performance
than a single decision tree. They are useful for classifying points with
discrete labels (i.e. 0
, 1
, 2
). This implementation of the
RandomForest
class is not for regression (i.e. predicting continuous
values).
mlpack’s RandomForest
class offers configurability via template parameters and
runtime parameters. This is used to provide the additional API-compatible
ExtraTrees
class. To use ExtraTrees
, simply replace RandomForest
with
ExtraTrees
in any of the documentation below. (More
information…)
Simple usage example:
// Train a random forest on random numeric data and predict labels on test data:
// All data and labels are uniform random; 10 dimensional data, 5 classes.
// Replace with a data::Load() call or similar for a real application.
arma::mat dataset(10, 1000, arma::fill::randu); // 1000 points.
arma::Row<size_t> labels =
arma::randi<arma::Row<size_t>>(1000, arma::distr_param(0, 4));
arma::mat testData(10, 500, arma::fill::randu); // 500 test points.
mlpack::RandomForest rf; // Step 1: create model.
rf.Train(dataset, labels, 5, 10); // Step 2: train model.
arma::Row<size_t> predictions;
rf.Classify(testData, predictions); // Step 3: classify points.
// You can also use `ExtraTrees` instead of `RandomForest`!
// Print some information about the test predictions.
std::cout << arma::accu(predictions == 3) << " test points classified as class "
<< "3." << std::endl;
Quick links:
- Constructors: create
RandomForest
objects. Train()
: train model.Classify()
: classify with a trained model.- Other functionality for loading, saving, and inspecting.
- Examples of simple usage and links to detailed example projects.
- Template parameters for custom behavior.
See also:
DecisionTree
DecisionTreeRegressor
- mlpack classifiers
- Random forest on Wikipedia
- Decision tree on Wikipedia
- Leo Breiman’s Random Forests page
🔗 Constructors
rf = RandomForest()
- Initialize the random forest without training.
- You will need to call
Train()
later to train the tree before callingClassify()
.
rf = RandomForest(data, labels, numClasses, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0)
rf = RandomForest(data, labels, numClasses, weights, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0)
- Train on numerical-only data (optionally with instance weights).
rf = RandomForest(data, info, labels, numClasses, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0)
rf = RandomForest(data, info, labels, numClasses, weights, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0)
- Train on mixed categorical data (optionally with instance weights).
Constructor Parameters:
name | type | description | default |
---|---|---|---|
data |
arma::mat |
Column-major training matrix. | (N/A) |
info |
data::DatasetInfo |
Dataset information, specifying type information for each dimension. | (N/A) |
labels |
arma::Row<size_t> |
Training labels, between 0 and numClasses - 1 (inclusive). Should have length data.n_cols . |
(N/A) |
numClasses |
size_t |
Number of classes in the dataset. | (N/A) |
weights |
arma::rowvec |
Instance weights for each training point. Should have length data.n_cols . |
(N/A) |
numTrees |
size_t |
Number of trees to train in the random forest. | 20 |
 |  |  |  |
minLeafSize |
size_t |
Minimum number of points in each leaf node of each decision tree. | 1 |
minGainSplit |
double |
Minimum gain for a node to split in each decision tree. | 1e-7 |
maxDepth |
size_t |
Maximum depth for each decision tree. (0 means no limit.) | 0 |
warmStart |
bool |
(Only available in Train() .) If true, training adds numTrees trees to the random forest. If false , an entirely new random forest will be created. |
false |
- If OpenMP is enabled, one thread will be used to train
each of the
numTrees
trees in the random forest. The computational effort involved with training a random forest increases linearly with the number of trees. - The default
minLeafSize
is1
, unlikeDecisionTree
. This is because random forests are less susceptible to overfitting due to their ensembled nature. - Note that the default
minLeafSize
of1
will make large decision trees, and so if a smaller-sized model is desired, this value should be increased (at the potential cost of accuracy). minGainSplit
can also be increased if a smaller-sized model is desired.
Note: different types can be used for data
and weights
(e.g.,
arma::fmat
, arma::sp_mat
). However, the element type of data
and
weights
must match; for example, if data
has type arma::fmat
, then
weights
must have type arma::frowvec
.
🔗 Training
If training is not done as part of the constructor call, it can be done with one
of the following versions of the Train()
member function:
rf.Train(data, labels, numClasses, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0, warmStart=false)
rf.Train(data, labels, numClasses, weights, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0, warmStart=false)
- Train on numerical-only data (optionally with instance weights).
- Returns a
double
with the average gain of each tree in the random forest. By default, this is the Gini gain, unless a differentFitnessFunction
template parameter is specified.
rf.Train(data, info, labels, numClasses, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0, warmStart=false)
rf.Train(data, info, labels, numClasses, weights, numTrees=20, minLeafSize=1, minGainSplit=1e-7, maxDepth=0, warmStart=false)
- Train on mixed categorical data (optionally with instance weights).
Types of each argument are the same as in the table for constructors above.
Notes:
-
The
warmStart
option, which allows incremental training (i.e. additional training on top of an existing model) is of typebool
and defaults tofalse
. This option is not available in the constructors. -
Train()
returns adouble
with the average gain of each tree in the random forest. By default, this is the Gini gain, unless a differentFitnessFunction
template parameter is specified.
🔗 Classification
Once a RandomForest
is trained, the Classify()
member function can be used
to make class predictions for new data.
size_t predictedClass = rf.Classify(point)
- (Single-point)
- Classify a single point, returning the predicted class.
rf.Classify(point, prediction, probabilitiesVec)
- (Single-point)
- Classify a single point and compute class probabilities.
- The predicted class is stored in
prediction
. - The probability of class
i
can be accessed withprobabilitiesVec[i]
.
rf.Classify(data, predictions)
- (Multi-point)
- Classify a set of points.
- The prediction for data point
i
can be accessed withpredictions[i]
.
rf.Classify(data, predictions, probabilities)
- (Multi-point)
- Classify a set of points and compute class probabilities for each point.
- The prediction for data point
i
can be accessed withpredictions[i]
. - The probability of class
j
for data pointi
can be accessed withprobabilities(j, i)
.
Classification Parameters:
usage | name | type | description |
---|---|---|---|
single-point | point |
arma::vec |
Single point for classification. |
single-point | prediction |
size_t& |
size_t to store class prediction into. |
single-point | probabilitiesVec |
arma::vec& |
arma::vec& to store class probabilities into. Will be set to length numClasses . |
 |  |  |  |
multi-point | data |
arma::mat |
Set of column-major points for classification. |
multi-point | predictions |
arma::Row<size_t>& |
Vector of size_t s to store class prediction into. Will be set to length data.n_cols . |
multi-point | probabilities |
arma::mat& |
Matrix to store class probabilities into (number of rows will be equal to number of classes, number of columns will be equal to data.n_cols ). |
Note: different types can be used for data
and point
(e.g.
arma::fmat
, arma::sp_mat
, arma::sp_vec
, etc.). However, the element type
that is used should be the same type that was used for training.
🔗 Other Functionality
-
A
RandomForest
can be serialized withdata::Save()
anddata::Load()
. -
rf.NumTrees()
will return asize_t
indicating the number of trees in the random forest. -
rf.Tree(i)
will return aDecisionTree
object representing thei
th decision tree in the random forest.
For complete functionality, the source code can be consulted. Each method is fully documented.
🔗 Simple Examples
See also the simple usage example for a trivial use of
RandomForest
.
Train a random forest incrementally on random mixed categorical data and save it to disk:
// Load a categorical dataset.
arma::mat dataset;
mlpack::data::DatasetInfo info;
// See https://datasets.mlpack.org/covertype.train.arff.
mlpack::data::Load("covertype.train.arff", dataset, info, true);
arma::Row<size_t> labels;
// See https://datasets.mlpack.org/covertype.train.labels.csv.
mlpack::data::Load("covertype.train.labels.csv", labels, true);
// Create the random forest.
mlpack::RandomForest rf;
// Train 10 trees on the given dataset, with a minimum leaf size of 3.
rf.Train(dataset, info, labels, 7 /* classes */, 10 /* trees */,
3 /* minimum leaf size */);
// Now load categorical test data.
arma::mat testDataset;
// See https://datasets.mlpack.org/covertype.test.arff.
mlpack::data::Load("covertype.test.arff", testDataset, info, true);
arma::Row<size_t> testLabels;
// See https://datasets.mlpack.org/covertype.test.labels.csv.
mlpack::data::Load("covertype.test.labels.csv", testLabels, true);
// Compute test set accuracy.
arma::Row<size_t> testPredictions;
rf.Classify(testDataset, testPredictions);
double accuracy = 100.0 * ((double) arma::accu(testPredictions == testLabels)) /
testLabels.n_elem;
std::cout << "After training 10 trees, test set accuracy is " << accuracy
<< "%." << std::endl;
// Now train another 10 trees and compute the test accuracy.
rf.Train(dataset, info, labels, 7 /* classes */, 10 /* trees */,
3 /* minimum leaf size */, 0.0 /* minimum split gain */,
0 /* maximum depth (unlimited) */, true /* incremental training */);
rf.Classify(testDataset, testPredictions);
accuracy = 100.0 * ((double) arma::accu(testPredictions == testLabels)) /
testLabels.n_elem;
std::cout << "After training 20 trees, test set accuracy is " << accuracy
<< "%." << std::endl;
// Save the random forest to disk.
mlpack::data::Save("rf.bin", "rf", rf);
Load a random forest and print some information about it.
mlpack::RandomForest rf;
// This call assumes a random forest called "rf" has already been saved to
// `rf.bin` with `data::Save()`.
mlpack::data::Load("rf.bin", "rf", rf, true);
std::cout << "The random forest in 'rf.bin' contains " << rf.NumTrees()
<< " trees." << std::endl;
if (rf.NumTrees() > 0)
{
std::cout << "The first tree's root node has " << rf.Tree(0).NumChildren()
<< " children." << std::endl;
}
Train a random forest on categorical data, and compare its performance with the performance of each individual tree:
// Load a categorical dataset (training and test sets).
arma::mat dataset, testDataset;
mlpack::data::DatasetInfo info;
arma::Row<size_t> labels, testLabels;
// See the following files:
// * https://datasets.mlpack.org/covertype.train.arff
// * https://datasets.mlpack.org/covertype.train.labels.csv
// * https://datasets.mlpack.org/covertype.test.arff
// * https://datasets.mlpack.org/covertype.test.labels.csv
mlpack::data::Load("covertype.train.arff", dataset, info, true);
mlpack::data::Load("covertype.train.labels.csv", labels, true);
mlpack::data::Load("covertype.test.arff", testDataset, info, true);
mlpack::data::Load("covertype.test.labels.csv", testLabels, true);
// Create the random forest.
mlpack::RandomForest rf;
// Train 20 trees on the given dataset, with a minimum leaf size of 5.
rf.Train(dataset, info, labels, 7 /* classes */, 20 /* trees */,
5 /* minimum leaf size */);
// Compute test set accuracy for each tree.
arma::Row<size_t> testPredictions;
for (size_t i = 0; i < rf.NumTrees(); ++i)
{
rf.Tree(i).Classify(testDataset, testPredictions);
const double accuracy = 100.0 *
((double) arma::accu(testPredictions == testLabels)) / testLabels.n_elem;
std::cout << "Tree " << i << " has test accuracy " << accuracy << "%."
<< std::endl;
}
// Now compute accuracy using the whole forest.
rf.Classify(testDataset, testPredictions);
const double accuracy = 100.0 *
((double) arma::accu(testPredictions == testLabels)) / testLabels.n_elem;
std::cout << "The whole forest has test accuracy " << accuracy << "%."
<< std::endl;
Train an ExtraTrees
model on random numeric data.
// 1000 random points in 10 dimensions.
arma::mat dataset(10, 1000, arma::fill::randu);
// Random labels for each point, totaling 5 classes.
arma::Row<size_t> labels =
arma::randi<arma::Row<size_t>>(1000, arma::distr_param(0, 4));
// Train in the constructor, using 10 trees in the forest.
// Note that `ExtraTrees` has exactly the same API as `RandomForest`.
mlpack::ExtraTrees<> rf(dataset, labels, 5, 10);
// Create a single test point.
arma::vec testPoint(10, arma::fill::randu);
size_t prediction;
arma::vec probabilities;
rf.Classify(testPoint, prediction, probabilities);
std::cout << "Test point predicted to be class " << prediction << "."
<< std::endl;
std::cout << "Probabilities of each class: " << probabilities.t();
See also the following fully-working examples:
🔗 Advanced Functionality: Template Parameters
Using different element types.
RandomForest
’s constructors, Train()
, and Predict()
functions support any
data type, so long as it supports the Armadillo matrix API. So, for instance,
learning can be done on single-precision floating-point data:
// 1000 random points in 10 dimensions.
arma::fmat dataset(10, 1000, arma::fill::randu);
// Random labels for each point, totaling 5 classes.
arma::Row<size_t> labels =
arma::randi<arma::Row<size_t>>(1000, arma::distr_param(0, 4));
// Train in the constructor.
mlpack::RandomForest rf(dataset, labels, 5);
// Create test data (500 points).
arma::fmat testDataset(10, 500, arma::fill::randu);
arma::Row<size_t> predictions;
rf.Classify(testDataset, predictions);
// Now `predictions` holds predictions for the test dataset.
// Print some information about the test predictions.
std::cout << arma::accu(predictions == 0) << " test points classified as class "
<< "0." << std::endl;
Fully custom behavior.
mlpack provides a few variants of the random forest classifier, using the
template parameters of the RandomForest
class. The following types can be
used as drop-in replacements throughout this documentation page:
RandomForest
- This is an implementation of Breiman’s seminal random forest algorithm (website, paper pdf).
- The
DecisionTree
class is used for each individual decision tree. - When training each individual decision tree, bootstrapping is used to compute the samples given to each tree for training.
ExtraTrees
- This is an implementation of the Extremely Randomized Trees algorithm (paper pdf).
- When training an
ExtraTrees
model, each individual decision tree chooses splits for numeric data randomly. - Training an
ExtraTrees
model is generally much faster thanRandomForest
, but the accuracy of theExtraTrees
model will be lower. - To use
ExtraTrees
, simply replaceRandomForest
withExtraTrees
in the documentation below.
Fully custom classes can also be used to control the behavior of the
RandomForest
class. The full signature of the class is as follows:
RandomForest<FitnessFunction,
DimensionSelectionType,
NumericSplitType,
CategoricalSplitType,
UseBootstrap>
FitnessFunction
: the measure of goodness to use when deciding on tree splitsDimensionSelectionType
: the strategy used for proposing dimensions to attempt to split onNumericSplitType
: the strategy used for finding splits on numeric data dimensionsCategoricalSplitType
: the strategy used for finding splits on categorical data dimensionsUseBootstrap
: a boolean indicating whether or not to use a bootstrap sample when training each tree in the forest
Note that the first four of these template parameters are exactly the same as
the template parameters for the
DecisionTree
class.
Below, details are given for the requirements of each of these template types.
FitnessFunction
- Specifies the fitness function to use when learning a decision tree.
- The
GiniGain
(default) andInformationGain
classes are available for drop-in usage. - A custom class must implement three functions:
// You can use this as a starting point for implementation.
class CustomFitnessFunction
{
// Return the range (difference between maximum and minimum gain values).
double Range(const size_t numClasses);
// Compute the gain for the given vector of labels, where `labels[i]` has an
// associated instance weight `weights[i]`.
//
// `RowType` and `WeightVecType` will be vector types following the Armadillo
// API. If `UseWeights` is `false`, then the `weights` vector should be
// ignored (e.g. the labels are not weighted).
template<bool UseWeights, typename RowType, typename WeightVecType>
double Evaluate(const RowType& labels,
const size_t numClasses,
const WeightVecType& weights);
// Compute the gain for the given counted set of labels, where `counts[i]`
// contains the number of points with label `i`. There are `totalCount`
// labels total, and `counts` has length `numClasses`.
//
// `UseWeights` is ignored, and `CountType` will be an integral type (e.g.
// `size_t`).
template<bool UseWeights, typename CountType>
double EvaluatePtr(const CountType* counts,
const size_t numClasses,
const CountType totalCount);
};
DimensionSelectionType
- When splitting a tree in the forest,
DimensionSelectionType
proposes possible dimensions to try splitting on. MultipleRandomDimensionSelect
(default) is available for drop-in usage and proposes a different random subset of dimensions at each decision tree node.- By default each random subset is of size
sqrt(d)
whered
is the number of dimensions in the data. - If constructed as
MultipleRandomDimensionSelect(n)
and passed to the constructor ofRandomForest
or theTrain()
function, each random subset will be of sizen
.
- By default each random subset is of size
- Each
RandomForest
constructor and each version of theTrain()
function optionally accept an instantiatedDimensionSelectionType
object as the very last parameter (aftermaxDepth
in the constructor, orwarmStart
inTrain()
), in case some internal state in the dimension selection mechanism is required. - A custom class must implement three simple functions:
class CustomDimensionSelect
{
public:
// Get the first dimension to try.
// This should return a value between `0` and `data.n_rows`.
size_t Begin();
// Get the next dimension to try. Note that internal state can be used to
// track which candidate dimension is currently being looked at.
// This should return a value between `0` and `data.n_rows`.
size_t Next();
// Get a value indicating that all dimensions have been tried.
size_t End() const;
// The usage pattern of `DimensionSelectionType` by `DecisionTree` is as
// follows, assuming that `dim` is an instantiated `DimensionSelectionType`
// object:
//
// for (size_t dim = dim.Begin(); dim != dim.End(); dim = dim.Next())
// {
// // ... try to split on dimension `dim` ...
// }
};
NumericSplitType
- Specifies the strategy to be used during training when splitting a numeric feature.
- The
BestBinaryNumericSplit
(default) class is available for drop-in usage and finds the best binary (two-way) split among all possible binary splits. - The
RandomBinaryNumericSplit
class is available for drop-in usage and will select a split randomly between the minimum and maximum values of a dimension. It is very efficient but does not yield splits that maximize the gain. (Used by theExtraTrees
variant.) - A custom class must take a
FitnessFunction
as a template parameter, implement three functions, and have an internal structureAuxiliarySplitInfo
that is used at classification time:
template<typename FitnessFunction>
class CustomNumericSplit
{
public:
// If a split with better resulting gain than `bestGain` is found, then
// information about the new, better split should be stored in `splitInfo` and
// `aux`. Specifically, a split is better than `bestGain` if the sum of the
// gains that the children will have (call this `sumChildrenGains`) is
// sufficiently better than the gain of the unsplit node (call this
// `unsplitGain`):
//
// split if `sumChildrenGains - unsplitGain > bestGain`, and
// `sumChildrenGains - unsplitGain > minGainSplit`, and
// each child will have at least `minLeafSize` points
//
// The new best split value should be returned (or anything greater than or
// equal to `bestGain` if no better split is found).
//
// If a new best split is found, then `splitInfo` and `aux` should be
// populated with the information that will be needed for
// `CalculateDirection()` to successfully choose the child for a given point.
// `splitInfo` should be set to a vector of length 1. The format of `aux` is
// arbitrary and is detailed more below.
//
// If `UseWeights` is false, the vector `weights` should be ignored.
// Otherwise, they are instance weighs for each value in `data` (one dimension
// of the input data).
template<bool UseWeights, typename VecType, typename WeightVecType>
double SplitIfBetter(const double bestGain,
const VecType& data,
const arma::Row<size_t>& labels,
const size_t numClasses,
const WeightVecType& weights,
const size_t minLeafSize,
const double minGainSplit,
arma::vec& splitInfo,
AuxiliarySplitInfo& aux);
// Return the number of children for a given split (stored as the single
// element from `splitInfo` and auxiliary data `aux` in `SplitIfBetter()`).
size_t NumChildren(const double& splitInfo,
const AuxiliarySplitInfo& aux);
// Given a point with value `point`, and split information `splitInfo` and
// `aux`, return the index of the child that corresponds to the point. So,
// e.g., if the split type was a binary split on the value `splitInfo`, you
// might return `0` if `point < splitInfo`, and `1` otherwise.
template<typename ElemType>
static size_t CalculateDirection(
const ElemType& point,
const double& splitInfo,
const AuxiliarySplitInfo& /* aux */);
// This class can hold any extra data that is necessary to encode a split. It
// should only be non-empty if a single `double` value cannot be used to hold
// the information corresponding to a split.
class AuxiliarySplitInfo { };
};
CategoricalSplitType
- Specifies the strategy to be used during training when splitting a categorical feature.
- The
AllCategoricalSplit
(default) is available for drop-in usage and splits all categories into their own node. - A custom class must take a
FitnessFunction
as a template parameter, implement three functions, and have an internal structureAuxiliarySplitInfo
that is used at classification time:
template<typename FitnessFunction>
class CustomCategoricalSplit
{
public:
// If a split with better resulting gain than `bestGain` is found, then
// information about the new, better split should be stored in `splitInfo` and
// `aux`. Specifically, a split is better than `bestGain` if the sum of the
// gains that the children will have (call this `sumChildrenGains`) is
// sufficiently better than the gain of the unsplit node (call this
// `unsplitGain`):
//
// split if `sumChildrenGains - unsplitGain > bestGain`, and
// `sumChildrenGains - unsplitGain > minGainSplit`, and
// each child will have at least `minLeafSize` points
//
// The new best split value should be returned (or anything greater than or
// equal to `bestGain` if no better split is found).
//
// If a new best split is found, then `splitInfo` and `aux` should be
// populated with the information that will be needed for
// `CalculateDirection()` to successfully choose the child for a given point.
// `splitInfo` should be set to a vector of length 1. The format of `aux` is
// arbitrary and is detailed more below.
//
// If `UseWeights` is false, the vector `weights` should be ignored.
// Otherwise, they are instance weighs for each value in `data` (one
// categorical dimension of the input data, which takes values between `0` and
// `numCategories - 1`).
template<bool UseWeights, typename VecType, typename LabelsType,
typename WeightVecType>
static double SplitIfBetter(
const double bestGain,
const VecType& data,
const size_t numCategories,
const LabelsType& labels,
const size_t numClasses,
const WeightVecType& weights,
const size_t minLeafSize,
const double minGainSplit,
arma::vec& splitInfo,
AuxiliarySplitInfo& aux);
// Return the number of children for a given split (stored as the single
// element from `splitInfo` and auxiliary data `aux` in `SplitIfBetter()`).
size_t NumChildren(const double& splitInfo,
const AuxiliarySplitInfo& aux);
// Given a point with (categorical) value `point`, and split information
// `splitInfo` and `aux`, return the index of the child that corresponds to
// the point. So, e.g., for `AllCategoricalSplit`, which splits a categorical
// dimension into one child for each category, this simply returns `point`.
template<typename ElemType>
static size_t CalculateDirection(
const ElemType& point,
const double& splitInfo,
const AuxiliarySplitInfo& /* aux */);
// This class can hold any extra data that is necessary to encode a split. It
// should only be non-empty if a single `double` value cannot be used to hold
// the information corresponding to a split.
class AuxiliarySplitInfo { };
};
UseBootstrap
- A
bool
value that indicates whether or not a bootstrap sample of the dataset should be used for the training of each individual decision tree in the random forest. - If
true
(default), a different bootstrap sample of the same size as the dataset will be used to train each decision tree. - If
false
(default for theExtraTrees
variant), the full dataset will be used to train each decision tree.