keel.Algorithms.Preprocess.Missing_Values.LLSImpute

Class LLSImpute

java.lang.Object

keel.Algorithms.Preprocess.Missing_Values.LLSImpute.LLSImpute

public class LLSImpute
extends java.lang.Object

This class implements the Local Least Squares Imputation

Author:

Julian Luengo Martin

Constructor Summary

Constructors

Constructor and Description
LLSImpute() Default constructor.
LLSImpute(java.lang.String fileParam) Constructor which extract the parameters from a KEEL pattern file and initializes the InstanceSet structures

Method Summary

Modifier and Type	Method and Description
protected void	config_read(java.lang.String fileParam) Parse the paramete file in KEEL format to obtain the parameters and working files
protected void	data2string(DenseMatrix mat, java.lang.String[][] X, InstanceSet IS) Parse the DenseMatrix of INPUT real values to a String 2D array, ready for printing to a file.
DenseMatrix	impute_llsq_l2_blind(DenseMatrix train, InstanceSet IS) Function that applies the Local Least Squares Imputation to a given array
DenseMatrix	impute_rowavg(DenseMatrix miss_matrix, int minexp, InstanceSet IS) Perform the row-average of given matrix
static DenseMatrix	inv(DenseMatrix A) Computes the inverse of a square non-singular Matrix
static DenseMatrix	pinv(DenseMatrix A) Computes the pseudoinverse of matrix A -> pinv(A) = V * pinv(S) * U' That is, Moore-Penrose pseudoinverse of a matrix If A is square and not singular, then pinv(A) is an expensive way to compute inv(A)
DenseMatrix	repmat(DenseMatrix mat, int m, int n) Replicate and tile an array.
void	run() Function that runs the LLSImpute over the data sets given in the pattern file in KEEL format
void	similargene(int i, java.util.ArrayList<java.lang.Integer> missidxj, java.util.ArrayList<java.lang.Integer> nomissidxj, int m, int n, DenseMatrix miss_matrix, DenseMatrix E, java.util.ArrayList<java.lang.Integer> gene_include, int m_gene_include) Computes the most similar (nearest) instances to a given one.
double	sum(DenseVector v) Compute the sum of all members of vector v
DenseVector	sumbyRows(DenseMatrix mat) From a given matrix mat, it performs the summatory by rows of such matrix
protected void	write_results(java.lang.String output, java.lang.String[][] X, InstanceSet IS) Write data matrix X to disk, in KEEL format

Methods inherited from class java.lang.Object

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

Constructor Detail

LLSImpute

public LLSImpute()

Default constructor.

LLSImpute

public LLSImpute(java.lang.String fileParam)

Constructor which extract the parameters from a KEEL pattern file and initializes the InstanceSet structures

Parameters:

fileParam -

Method Detail

config_read

protected void config_read(java.lang.String fileParam)

Parse the paramete file in KEEL format to obtain the parameters and working files

Parameters:

fileParam - Pattern file in KEEL format

run

public void run()

Function that runs the LLSImpute over the data sets given in the pattern file in KEEL format

impute_llsq_l2_blind

public DenseMatrix impute_llsq_l2_blind(DenseMatrix train,
                                        InstanceSet IS)

Function that applies the Local Least Squares Imputation to a given array

Parameters:

train - The Matrix with the data

IS - The original Instance set, used for reference (attributes, etc.)

Returns:

A new allocated DenseMatrix of same size than 'train' which has the missing values imputed

similargene

public void similargene(int i,
                        java.util.ArrayList<java.lang.Integer> missidxj,
                        java.util.ArrayList<java.lang.Integer> nomissidxj,
                        int m,
                        int n,
                        DenseMatrix miss_matrix,
                        DenseMatrix E,
                        java.util.ArrayList<java.lang.Integer> gene_include,
                        int m_gene_include)

Computes the most similar (nearest) instances to a given one. The result are stoed in matrix A,B and vector w, which are fields of the current object.

Parameters:

i - The number of the given instance in the E matrix (i.e. the row number)

missidxj - Indices of the missing attributes

nomissidxj - Indices of the non-missing (complete) attributes

m - Rows of the E matrix

n - Columns of the E matrix

miss_matrix - Original matrix which has no missing value imputed, same size of E

E - Working matrix which has the previous found missing values estimated

gene_include - The indices of the genes (rows or instances) which we will consider as neighbours to instance i

m_gene_include - Number of neighbours to be considered (size of gene_include)

impute_rowavg

public DenseMatrix impute_rowavg(DenseMatrix miss_matrix,
                                 int minexp,
                                 InstanceSet IS)

Perform the row-average of given matrix

Parameters:

miss_matrix - The original matrix with all missing values

minexp - The minimum number of non-missing values to compute the row-average

IS - The reference InstanceSet

Returns:

the row-average dense matrix.

pinv

public static DenseMatrix pinv(DenseMatrix A)

Computes the pseudoinverse of matrix A -> pinv(A) = V * pinv(S) * U' That is, Moore-Penrose pseudoinverse of a matrix If A is square and not singular, then pinv(A) is an expensive way to compute inv(A)

Parameters:

A - The matrix from we compute the pseudoinverse

Returns:

The pseudoinverse of matrix A

inv

public static DenseMatrix inv(DenseMatrix A)

Computes the inverse of a square non-singular Matrix

Parameters:

A - The Matrix from which we will compute the inverse

Returns:

A new allocated matrix which contains the inverse of A

sum

public double sum(DenseVector v)

Compute the sum of all members of vector v

Parameters:

v - The reference vector

Returns:

Summatory of all elements of v

sumbyRows

public DenseVector sumbyRows(DenseMatrix mat)

From a given matrix mat, it performs the summatory by rows of such matrix

Parameters:

mat - The reference matrix

Returns:

A new allocated array with the sum of each row's elements

repmat

public DenseMatrix repmat(DenseMatrix mat,
                          int m,
                          int n)

Replicate and tile an array. B = repmat(A,m,n) creates a large matrix B consisting of an m-by-n tiling of copies of A. The size of B is [size(A,1)*m, (size(A,2)*n]. The statement repmat(A,n) creates an n-by-n tiling.

Parameters:

mat - The original matrix

m - Number of rowwise replications

n - Number of columnwise replications

Returns:

A new allocated matrix with the tiling of matrix mat

data2string

protected void data2string(DenseMatrix mat,
                           java.lang.String[][] X,
                           InstanceSet IS)

Parse the DenseMatrix of INPUT real values to a String 2D array, ready for printing to a file. It also fits the values to the original bounds if needed.

Parameters:

mat - The DenseMatrix with the input values in double format

X - The output String matrix, ready to be printed

IS - The InstanceSet with the original values, used to obtain the OUTPUT values

write_results

protected void write_results(java.lang.String output,
                             java.lang.String[][] X,
                             InstanceSet IS)

Write data matrix X to disk, in KEEL format

Parameters:

output - The file to which we print

X - The 2D array with the values of the attributes parsed to a string

IS - The reference InstanceSet