jsat.linear

Class MatrixStatistics

java.lang.Object

jsat.linear.MatrixStatistics

public class MatrixStatistics
extends java.lang.Object

This class provides methods useful for statistical operations that involve matrices and vectors.

Method Summary

Modifier and Type	Method and Description
static Vec	covarianceDiag(Vec means, DataSet dataset) Computes the weighted diagonal of the covariance matrix, which is the standard deviations of the columns of all values.
static <V extends Vec> Vec	covarianceDiag(Vec means, java.util.List<V> dataset) Computes the diagonal of the covariance matrix, which is the standard deviations of the columns of all values.
static void	covarianceDiag(Vec means, Vec diag, DataSet dataset) Computes the weighted diagonal of the covariance matrix, which is the standard deviations of the columns of all values.
static <V extends Vec> void	covarianceDiag(Vec means, Vec diag, java.util.List<V> dataset) Computes the diagonal of the covariance matrix, which is the standard deviations of the columns of all values.
static Matrix	covarianceMatrix(Vec mean, DataSet dataSet) Computes the weighted covariance matrix of the data set
static void	covarianceMatrix(Vec mean, DataSet dataSet, Matrix covariance) Computes the weighted covariance matrix of the given data set.
static void	covarianceMatrix(Vec mean, DataSet dataSet, Matrix covariance, double sumOfWeights, double sumOfSquaredWeights) Computes the weighted covariance matrix of the given data set.
static void	covarianceMatrix(Vec mean, java.util.List<DataPoint> dataSet, Matrix covariance) Computes the weighted result for the covariance matrix of the given data set.
static void	covarianceMatrix(Vec mean, java.util.List<DataPoint> dataSet, Matrix covariance, double sumOfWeights, double sumOfSquaredWeights) Computes the weighted result for the covariance matrix of the given data set.
static <V extends Vec> Matrix	covarianceMatrix(Vec mean, java.util.List<V> dataSet)
static <V extends Vec> void	covarianceMatrix(Vec mean, Matrix covariance, java.util.List<V> dataSet)
static <V extends Vec> void	covarianceMatrix(Vec mean, Matrix covariance, java.util.List<V> dataSet, java.util.Collection<java.lang.Integer> subset)
static <V extends Vec> void	FastMCD(Vec mean, Matrix cov, java.util.List<V> dataset, boolean parallel) This algorithm implements the FastMCD algorithm for robustly estimating the mean and covariance of a dataset.
static Vec	meanVector(DataSet dataSet) Computes the weighted mean of the given data set.
static <V extends Vec> Vec	meanVector(java.util.List<V> dataSet) Computes the mean of the given data set.
static void	meanVector(Vec mean, DataSet dataSet) Computes the weighted mean of the data set
static <V extends Vec> void	meanVector(Vec mean, java.util.List<V> dataSet) Computes the mean of the given data set.
static <V extends Vec> void	meanVector(Vec mean, java.util.List<V> dataSet, java.util.Collection<java.lang.Integer> subset) Computes the mean of the given data set.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

meanVector

public static <V extends Vec> Vec meanVector(java.util.List<V> dataSet)

Computes the mean of the given data set.

Type Parameters:

V - the vector type

Parameters:

dataSet - the list of vectors to compute the mean of

Returns:

the mean of the vectors

meanVector

public static Vec meanVector(DataSet dataSet)

Computes the weighted mean of the given data set.

Parameters:

dataSet - the dataset to compute the mean from

Returns:

the mean of the numeric vectors in the data set

meanVector

public static <V extends Vec> void meanVector(Vec mean,
                                              java.util.List<V> dataSet)

Computes the mean of the given data set.

Parameters:

mean - the zeroed out vector to store the mean in. Its contents will be altered

dataSet - the set of data points to compute the mean from

meanVector

public static <V extends Vec> void meanVector(Vec mean,
                                              java.util.List<V> dataSet,
                                              java.util.Collection<java.lang.Integer> subset)

Computes the mean of the given data set.

Type Parameters:

V -

Parameters:

mean - the zeroed out vector to store the mean in. Its contents will be altered

dataSet - the set of data points to compute the mean from

subset - the indecies of the points in dataSet to take the mean of

meanVector

public static void meanVector(Vec mean,
                              DataSet dataSet)

Computes the weighted mean of the data set

Parameters:

mean - the zeroed out vector to store the mean in. Its contents will be altered

dataSet - the set of data points to compute the mean from

covarianceMatrix

public static <V extends Vec> Matrix covarianceMatrix(Vec mean,
                                                      java.util.List<V> dataSet)

covarianceMatrix

public static <V extends Vec> void covarianceMatrix(Vec mean,
                                                    Matrix covariance,
                                                    java.util.List<V> dataSet)

covarianceMatrix

public static <V extends Vec> void covarianceMatrix(Vec mean,
                                                    Matrix covariance,
                                                    java.util.List<V> dataSet,
                                                    java.util.Collection<java.lang.Integer> subset)

covarianceMatrix

public static void covarianceMatrix(Vec mean,
                                    java.util.List<DataPoint> dataSet,
                                    Matrix covariance)

Computes the weighted result for the covariance matrix of the given data set. If all weights have the same value, the result will come out equivalent to covarianceMatrix(jsat.linear.Vec, java.util.List)

Parameters:

mean - the mean of the distribution.

dataSet - the set of data points that contain vectors

covariance - the zeroed matrix to store the result in. Its values will be altered.

covarianceMatrix

public static void covarianceMatrix(Vec mean,
                                    java.util.List<DataPoint> dataSet,
                                    Matrix covariance,
                                    double sumOfWeights,
                                    double sumOfSquaredWeights)

Computes the weighted result for the covariance matrix of the given data set. If all weights have the same value, the result will come out equivalent to covarianceMatrix(jsat.linear.Vec, java.util.List)

Parameters:

mean - the mean of the distribution.

dataSet - the set of data points that contain vectors

covariance - the zeroed matrix to store the result in. Its values will be altered.

sumOfWeights - the sum of each weight in dataSet

sumOfSquaredWeights - the sum of the squared weights in dataSet

covarianceMatrix

public static Matrix covarianceMatrix(Vec mean,
                                      DataSet dataSet)

Computes the weighted covariance matrix of the data set

Parameters:

mean - the mean of the data set

dataSet - the dataset to compute the covariance of

Returns:

the covariance matrix of the data set

covarianceMatrix

public static void covarianceMatrix(Vec mean,
                                    DataSet dataSet,
                                    Matrix covariance)

Computes the weighted covariance matrix of the given data set.

Parameters:

mean - the mean of the data set

dataSet - the dataset to compute the covariance of

covariance - the zeroed out matrix to store the result into

covarianceMatrix

public static void covarianceMatrix(Vec mean,
                                    DataSet dataSet,
                                    Matrix covariance,
                                    double sumOfWeights,
                                    double sumOfSquaredWeights)

Computes the weighted covariance matrix of the given data set. Superfluous calculations are avoided by having the call provide information.

Parameters:

mean - the mean of the data set

dataSet - the dataset to compute the covariance of

covariance - the zeroed out matrix to store the result into

sumOfWeights - the sum of the weights for each data point in the dataset

sumOfSquaredWeights - the sum of the squared weights for each data point in the data set

covarianceDiag

public static void covarianceDiag(Vec means,
                                  Vec diag,
                                  DataSet dataset)

Computes the weighted diagonal of the covariance matrix, which is the standard deviations of the columns of all values.

Parameters:

means - the already computed mean of the data set

diag - the zeroed out vector to store the diagonal in. Its contents will be altered

dataset - the data set to compute the covariance diagonal from

covarianceDiag

public static Vec covarianceDiag(Vec means,
                                 DataSet dataset)

Computes the weighted diagonal of the covariance matrix, which is the standard deviations of the columns of all values.

Parameters:

means - the already computed mean of the data set

dataset - the data set to compute the covariance diagonal from

Returns:

the diagonal of the covariance matrix for the given data

covarianceDiag

public static <V extends Vec> void covarianceDiag(Vec means,
                                                  Vec diag,
                                                  java.util.List<V> dataset)

Computes the diagonal of the covariance matrix, which is the standard deviations of the columns of all values.

Type Parameters:

V - the type of the vector

Parameters:

means - the already computed mean of the data set

diag - the zeroed out vector to store the diagonal in. Its contents will be altered

dataset - the data set to compute the covariance diagonal from

covarianceDiag

public static <V extends Vec> Vec covarianceDiag(Vec means,
                                                 java.util.List<V> dataset)

Computes the diagonal of the covariance matrix, which is the standard deviations of the columns of all values.

Type Parameters:

V -

Parameters:

means - the already computed mean of the data set

dataset - the data set to compute the covariance diagonal from

Returns:

the diagonal of the covariance matrix for the given data

FastMCD

public static <V extends Vec> void FastMCD(Vec mean,
                                           Matrix cov,
                                           java.util.List<V> dataset,
                                           boolean parallel)

This algorithm implements the FastMCD algorithm for robustly estimating the mean and covariance of a dataset. Computational complexity increases linearly with the sample size n, but cubically with the dimension size d.

See: Rousseeuw, P. J., & Driessen, K. Van. (1999). A Fast Algorithm for the Minimum Covariance Determinant Estimator. Technometrics, 41(3), 212–223. http://doi.org/10.2307/1270566

Type Parameters:

V -

Parameters:

mean - the location to store the estimated mean, values will be overwritten

cov - the location to store the estimated covariance, values will be overwritten

dataset - the set of data points to estimate the mean and covariance of

parallel - true if multiple cores should be used for estimation, false for single thread.