Faq how do manually calculate the limits for scores

From Eigenvector Research Documentation Wiki
Jump to navigation Jump to search

Issue:

How do I manually calculate the limits for scores?

Possible Solutions:

Individual scores limits can be calculated using the code below, given a PCA model (modl), the pc of interest (pc), and the % confidence limit level desired (limitsvalue):

>> pc = 2;  %which PC to calculate limit for
>> limitsvalue = 95;  % confidence level (in %)
>> L = modl.detail.ssq(pc,2);            %get eigenvalue
>> m = length(modl.detail.includ{1,1});  %get number of samples
>> scorelimit = sqrt(L * tsqlim(m,1,limitsvalue));

To calculate scores limits for PCR models, the above must be slightly modified because the eigenvalues are stored in a different field. Replace the L = ... line above with this:

>> L = modl.detail.pcassq(pc,2);         %get PCR model eigenvalue

For PLS models, you must approximate L (the eigenvalue) from the RMSE of the calibration data scores:

>> L = rmse(modl.loads{1,1}(:,pc)).^2;   %estimate PLS model eigenvalue

This code assumes that no samples were excluded in the model. If any samples were excluded, use the modl.detail.includ{1} field to calculate L from only the included samples.

Notes:

To calculate limits for T2 and Q, see the related FAQ.


Still having problems? Please contact our helpdesk at helpdesk@eigenvector.com