This is a list of the main routines you can use to build models by means of the PCA toolbox for MATLAB:
pca_gui opens a GUI for calculating all models provided in the toolbox; in order to open the graphical interface, type:
on the MATLAB command window. There are no inputs; data, labels and models can be loaded and saved directly from the graphical interface.
Routines for fitting models
Principal Component Analysis (PCA), Muiltidimensional Scaling (MDS) and Cluster Analysis can be calculated by means of the routines pca_model, mds_model and cluser_model respectively. The output of the routines collects the calculated results in a matlab structure. The pca_compsel routine performs and calculate different parameters to estimate the optimal number of components to be retained. Type "help routine_name" on the MATLAB command window for further information.
Routines for predicting new samples (for PCA)
Prediction of new samples by means of PCA can be performed with the routine pca_project. The output of the routine collects the scores of the predicted samples in the PCA model and other details (such as Q residuals and T2 Hotelling values). Type "help pca_project " on the MATLAB command window for further informations.