Plspulsm: Difference between revisions

From Eigenvector Research Documentation Wiki
Jump to navigation Jump to search
imported>Jeremy
(Importing text file)
imported>Jeremy
No edit summary
 
(2 intermediate revisions by one other user not shown)
Line 1: Line 1:
===Purpose===
===Purpose===


Builds finite impulse response (FIR) models for multi-input single (MISO) output systems using partial least squares regression.
Builds finite impulse response (FIR) models for multi-input single output (MISO) systems using partial least squares regression.


===Synopsis===
===Synopsis===
Line 10: Line 9:
===Description===
===Description===


plspulsm calculates a vector of FIR coefficients b using PLS regression. Inputs are a matrix of process input vectors u, and a process output vector y. n is a row vector with the number of FIR coefficents to use for each input, maxlv is the maximum number of latent variables to consider, split is the number of times the model is rebuilt and tested during cross-validation, and delay is a row vector containing the number of time units of delay for each input.
plspulsm calculates a vector of FIR coefficients <tt>b</tt> using PLS regression.  
 
'''Note''': '''plspulsm''' uses contiguous blocks of data for cross-validation.
 
====Inputs====
 
* '''u''' = matrix of process input vectors
* '''y''' = process output vector
* '''n''' = a row vector with the number of FIR coefficients to use for each input
* '''maxlv''' = maximum number of latent variables to consider
* '''split''' = number of times the model is rebuilt and tested during cross-validation
* '''delay''' = row vector containing the number of time units of delay for each input
 
====Outputs====


Note: plspulsm uses contiguous blocks of data for cross-validation.
* '''b''' = vector of FIR coefficients


===Examples===
===Examples===


b = plspulsm([u1 u2],y,[25 15],5,10,[0 3])
:b = plspulsm([u1 u2],y,[25 15],5,10,[0 3])


This system has 2 inputs as column vectors u1 and u2 and a single output vector y. The FIR model will use 25 coefficients for input variable u1 and 15 coefficients for input variable u2. For this model a maximum of 5 latent variables will be considered. The cross validation split the data into 10 subsets. The number of time units of delay for the first input variable u1 is 0 and for the second input variable u2 it is 3.  
In this example, the system has 2 inputs as column vectors u1 and u2 and a single output vector y. The FIR model will use 25 coefficients for input variable u1 and 15 coefficients for input variable u2. For this model a maximum of 5 latent variables will be considered. The cross validation split the data into 10 block-wise subsets. The number of time units of delay for the first input variable u1 is 0 and for the second input variable u2 it is 3.  


===See Also===
===See Also===


[[autocor]], [[crosscor]], [[fir2ss]], [[wrtpulse]]
[[autocor]], [[crosscor]], [[fir2ss]], [[wrtpulse]]

Latest revision as of 14:42, 10 October 2008

Purpose

Builds finite impulse response (FIR) models for multi-input single output (MISO) systems using partial least squares regression.

Synopsis

b = plspulsm(u,y,n,maxlv,split,delay)

Description

plspulsm calculates a vector of FIR coefficients b using PLS regression.

Note: plspulsm uses contiguous blocks of data for cross-validation.

Inputs

  • u = matrix of process input vectors
  • y = process output vector
  • n = a row vector with the number of FIR coefficients to use for each input
  • maxlv = maximum number of latent variables to consider
  • split = number of times the model is rebuilt and tested during cross-validation
  • delay = row vector containing the number of time units of delay for each input

Outputs

  • b = vector of FIR coefficients

Examples

b = plspulsm([u1 u2],y,[25 15],5,10,[0 3])

In this example, the system has 2 inputs as column vectors u1 and u2 and a single output vector y. The FIR model will use 25 coefficients for input variable u1 and 15 coefficients for input variable u2. For this model a maximum of 5 latent variables will be considered. The cross validation split the data into 10 block-wise subsets. The number of time units of delay for the first input variable u1 is 0 and for the second input variable u2 it is 3.

See Also

autocor, crosscor, fir2ss, wrtpulse