DLims

Location: ..\cubatch\Model_PARAFAC\private\DLims.m

function StatsOut = DLims(Pred_Fun,Mode,Limit,Data,Model,Validation,ConfLim,FillIn,Segments,StatsIn,Cen,Scal,PreProc);

Description:
Computes the D-statistics, values, its limits, the contribution to the D-statistic of the different variables in the various dimensions of the array for a given nPLS1/PLS1 model.
Apart from the contributions the statistics can be computed both in an
'offline' and an 'online' mode.
NB This function will likely undergo some simplifications in the future releases of the syntax as many of the input parameters are contained in the same data structure.


Inputs:
Pred_Fun: function computing the prediction for the desired model. It can be a string (for MatLab© 5.3 or superior) or  a function handle (for MatLab© 6.x or superior)
Mode: 'on'/'off'  =>  'online'/'offline'
Limit: 0/1 compute/do not compute D-statistics limits (based on Data for the D-statistics contributions)
Data: array of doubles with the data used to compute the statistics
Model
:
model's parameters as required by Pred_Fun (i.e. as a structure ModelOut.model)
Validation: model's parameters as determined throught the validation procedure (ModelOut.validation). It is not currently used but it can be employed in future releases to compute confidence limits for the contributions to the D-statistic.
ConfLim
: "alpha"s for the confidence intervals (if empty or not given: 95% and 99%).
FillIn: fill in method:
'Current deviation'/'zero', necessary only when Mode is 'on'
Segments
: Validation segments (ModelOut.validation.segments). Necessary only for D-statistics contributions.
NB. this parameter is somewhat superfluous as already contained in Validation and therefore will likely be eliminated in future versions
StatsIn
: structure holding the statistics (if previously calculated)
Cen,Scal
: preprocessing parameters as contained in Model.xpreproc.
PreProcess: Stores the preprocessing information for the model of reference.
It contains two fields: 'modx' (for X) and 'mody' (for Y). Each of them has two fields: 'cen' and 'scal', which are row vectors of the same length as the original X and Y and refer to centre and scaling.
1 means centre/scale according to mean and standard deviation.
The 'scal' values can also be 2 (scaling with Root Mean Squares) or 3 (scaling with Frobenius norm)

NB
  these parameters are somewhat superfluous as already contained in Model and therefore will likely be eliminated in future versions

Outputs:
StatsOut: structure holding the various statistics relative to Data. With respect to StatsIn the fields relative to the D-statistics (with respect to Mode) are filled in.

Called by:
Model_PARAFAC\Plot_DStatistics

Subroutines:
InternalNone
External: onlineres, Pred_Fun

Author:
Giorgio Tomasi
Royal Agricultural and Veterinary University
MLI, LMT, Chemometrics group
Rolighedsvej 30
DK-1958 Frederiksberg C
Danmark

Last modified: 04-Nov-2002 12:05:42

Contact: Giorgio Tomasi, gt@kvl.dk

References
[1] Nomikos P., Mac Gregor J.F., "Monitoring batch process using multiway principal component analysis", AIChE journal, Vol 40, n°8, 1994, 1361-1373
[2]
Westerhuis,J.A.; Gurden,S.P.; Smilde,A.K.,"Generalized contribution plots in multivariat statistical process monitoring", Chemometrics and Intelligent Laboratory Systems, Vol 51, 2000, 96-114