# Difference between revisions of "Distslct"

imported>Jeremy (Importing text file) |
imported>Donal (→See Also) |
||

(4 intermediate revisions by 3 users not shown) | |||

Line 1: | Line 1: | ||

===Purpose=== | ===Purpose=== | ||

+ | |||

Select samples on the exterior of a data space based on a Euclidean distance. | Select samples on the exterior of a data space based on a Euclidean distance. | ||

+ | |||

===Synopsis=== | ===Synopsis=== | ||

+ | |||

:isel = distslct(x,nosamps,''flag'') | :isel = distslct(x,nosamps,''flag'') | ||

+ | |||

===Description=== | ===Description=== | ||

− | DISTSLCT first identifies a sample in the ''M'' by ''N'' data set x furthest from the data set mean. Subsequent samples are selected to be simultaneously the furthest from the mean and the selected samples for a total of nosamps selected samples. DISTSLCT calls STDSSLCT to find the number of samples up to the rank of the data and uses a distance measure to find additional samples if nosamps>rank(x). | + | |

− | Optional intput tells DISTSLCT how many samples STDSLCT should estimate when nosamps>''N'': | + | DISTSLCT first identifies a sample in the ''M'' by ''N'' data set x furthest from the data set mean. Subsequent samples are selected to be simultaneously the furthest from the mean and the selected samples for a total of <tt>nosamps</tt> selected samples. DISTSLCT calls STDSSLCT to find the number of samples up to the rank of the data and uses a distance measure to find additional samples if <tt>nosamps>rank(x)</tt>. |

− | * 1 = STDSLCT | + | |

− | * 2 = STDSLCT selects ''N'' {default}. | + | Optional intput tells DISTSLCT how many samples STDSLCT should estimate when <tt>nosamps</tt>>''N'': |

− | Output isel is a vector of length nosamps containing the indices of the selected samples. | + | |

+ | * '''1''' = STDSLCT selects ''N''-1, or | ||

+ | |||

+ | * '''2''' = STDSLCT selects ''N'' {default}. | ||

+ | |||

+ | Output <tt>isel</tt> is a vector of length <tt>nosamps</tt> containing the indices of the selected samples. | ||

+ | |||

This routine is used to initialize the selection of samples in the DOPTIMAL function. Altough it does not satisfy the d-optimality condition, it is an alternative to doptimal that does not require an inverse or calculation of a determinant. | This routine is used to initialize the selection of samples in the DOPTIMAL function. Altough it does not satisfy the d-optimality condition, it is an alternative to doptimal that does not require an inverse or calculation of a determinant. | ||

+ | |||

+ | ====Inputs==== | ||

+ | |||

+ | * '''x''': data set, ''M'' by ''N'' | ||

+ | * '''nosamps''': number of selected samples | ||

+ | |||

+ | ====Optional Inputs==== | ||

+ | |||

+ | * '''flag''': how many samples to select when <tt>nosamps</tt>>''N''; a value of 1 selects ''N''-1, while a value of 2 (default) selects ''N''. | ||

+ | |||

+ | ====Outputs==== | ||

+ | |||

+ | * '''isel''': vector containing the indices of the selected samples | ||

+ | |||

===See Also=== | ===See Also=== | ||

− | [[doptimal]], [[stdsslct]] | + | |

+ | [[doptimal]], [[reducennsamples]], [[stdsslct]], [[splitcaltest]] |

## Latest revision as of 12:04, 22 March 2013

### Purpose

Select samples on the exterior of a data space based on a Euclidean distance.

### Synopsis

- isel = distslct(x,nosamps,
*flag*)

### Description

DISTSLCT first identifies a sample in the *M* by *N* data set x furthest from the data set mean. Subsequent samples are selected to be simultaneously the furthest from the mean and the selected samples for a total of `nosamps` selected samples. DISTSLCT calls STDSSLCT to find the number of samples up to the rank of the data and uses a distance measure to find additional samples if `nosamps>rank(x)`.

Optional intput tells DISTSLCT how many samples STDSLCT should estimate when `nosamps`>*N*:

**1**= STDSLCT selects*N*-1, or

**2**= STDSLCT selects*N*{default}.

Output `isel` is a vector of length `nosamps` containing the indices of the selected samples.

This routine is used to initialize the selection of samples in the DOPTIMAL function. Altough it does not satisfy the d-optimality condition, it is an alternative to doptimal that does not require an inverse or calculation of a determinant.

#### Inputs

**x**: data set,*M*by*N***nosamps**: number of selected samples

#### Optional Inputs

**flag**: how many samples to select when`nosamps`>*N*; a value of 1 selects*N*-1, while a value of 2 (default) selects*N*.

#### Outputs

**isel**: vector containing the indices of the selected samples