Osccalc: Difference between revisions
Jump to navigation
Jump to search
imported>Scott |
imported>Jeremy |
||
| (One intermediate revision by the same user not shown) | |||
| Line 13: | Line 13: | ||
Optional inputs are the maximum number of iterations used in attempting to maximize the variance captured by othogonal components iter {default = 0}, and the tolerance on percent of x variance to consider when forming the final w vector <tt>tol</tt> {default = 99.9}. | Optional inputs are the maximum number of iterations used in attempting to maximize the variance captured by othogonal components iter {default = 0}, and the tolerance on percent of x variance to consider when forming the final w vector <tt>tol</tt> {default = 99.9}. | ||
Outputs are the OSC corrected predictor matrix <tt>nx</tt>, and the x-block weigths <tt>nw< | Outputs are the OSC corrected predictor matrix <tt>nx</tt>, and the x-block weigths <tt>nw</tt>, loads <tt>np</tt>, and scores <tt>nt</tt> that were used in making the correction. | ||
Once the calibration is done, new (scaled) X data can be corrected by <tt>newx = x - x | Once the calibration is done, new (scaled) X data can be corrected by <tt>newx = x - x*nw*inv(np'*nw)*np';</tt> (See OSCAPP). | ||
===See Also=== | ===See Also=== | ||
[[caltransfer]], [[crossval]], [[glsw]], [[oscapp]] | [[caltransfer]], [[crossval]], [[glsw]], [[oscapp]] | ||
Latest revision as of 09:19, 16 March 2012
Purpose
Calculates orthogonal signal correction.
Synopsis
- [nx,nw,np,nt] = osccalc(x,y,nocomp,iter,tol)
Description
Inputs are the matrix of scaled predictor variables x, scaled predicted variable(s) y, and the number of OSC components nocomp.
Optional inputs are the maximum number of iterations used in attempting to maximize the variance captured by othogonal components iter {default = 0}, and the tolerance on percent of x variance to consider when forming the final w vector tol {default = 99.9}.
Outputs are the OSC corrected predictor matrix nx, and the x-block weigths nw, loads np, and scores nt that were used in making the correction.
Once the calibration is done, new (scaled) X data can be corrected by newx = x - x*nw*inv(np'*nw)*np'; (See OSCAPP).