What's new in version 7.0
Random Forest (RF) was added as classification method. Minor bugs were fixed.
[-> top]
What's new in version 6.0
Variance scaling and range scaling has been added as approaches for column pre-processing. Basic row pre-processing has been added (standard normal variate, multiplicatve scatter correction, first and second derivatives). Savitzky-Golay smoothing can now be applied for analytical spectral data. The layout of the calculation menu has been simplified.
[-> top]
What's new in version 5.4
Class modelling methods (SIMCA, Potential Functions, UNEQ) can now be calculated just on a specific target class. The layout of the form with classification measures have been updated.
[-> top]
What's new in version 5.3
Backpropagation Neural Networks (BPNN) was added as classification method. The maximum number of classes that can be loaded (and modelled) has been extended to 20.
[-> top]
What's new in version 5.2
The graphical interface has been improved. Now forms with plots can be resized. Classes can be loaded both as numerical or string arrays. Calculation of class probabilities was added in Discriminant Analysis and SVM routines.
[-> top]
What's new in version 5.1
Visualisation of class potential in the score plots of classification results was improved. Calculation of class probabilities was added in PLSDA routines.
[-> top]
What's new in version 5.0
Unequal class models (UNEQ) was added as classification method.
Visualisation of class potential has been added in the score plots of classification results (for methods which have such plots).
[-> top]
What's new in version 4.2
Minor bugs were fixed. Forms were modified in order to get the same dimension on different Matlab versions. Modules for Classification trees (CART) were changed according to new modifications of the Matlab statistic toolbox. Bootstrap and montecarlo validation procedures were introduced in the GUI.
[-> top]
What's new in version 4.1
In the toolbox version 4.1, it is possible to analyse class boundaries when dealing with the classification of 2-dimensional datasets or when models are calculated on two principal components / latent variables.
[-> top]
What's new in version 4.0
In the toolbox version 4.0, two classification methods have been added: Potential Functions (Kernel Density Estimators) and Support Vector Machines (SVM).
Models calculated on the MATLAB command window can now be imported in the graphical interface (for further graphical analysis or prediction on new samples).
When dealing with PLSDA and SIMCA, area under ROC curve (AUC) is calculated and shown in the ROC curve plots.
[-> top]
What's new in version 3.6
In the toolbox version 3.6, the class modelling criterion for classify samples with Soft Independent Modeling of Class Analogy (SIMCA) was changed, as well as the procedure for the selection of the optimal number of components.
Q residuals contribution and Hotelling T2 contributions are now calculated for SIMCA, PLSDA and PCA-LDA models. Values are available in the model structure (see Routines for fitting models) or can be visualised in the graphical interface.
A new menu item "view->plot profiles" has been added. This opens a new form to show the sample profiles and the profiles of the variable averages on the raw and scaled data.
The default number of cross validation groups in the GUI has been set to 10 for all classification methods.
[-> top]
What's new in version 3.1
In the toolbox version 3.1, classification models based on Soft Independent Modeling of Class Analogy (SIMCA) were added. A form for plotting univariate statistics of variables has been added in the "view" menu.
Samples can be deleted in the "view" menu. Prediction of external samples can be saved to the MATLAB workspace in the "file" menu. The projection of test samples in the model space can be analysed, if SIMCA, PLSDA or DA based on PCA were calculated.
When dealing with PLSDA, Q residuals and Hotelling T2 now are plotted also for test samples.
The calculation of non error rate and error rate for multi class problems has been modified. Details are here.
[-> top]
What's new in version 2.0
In the toolbox version 2.0, a GUI graphical interface has been added. The graphical interface enables you to do all the steps of the analysis (data loading, setting preparation, model calculation, sample prediction, cross-validation). If you wish to use the graphical interface, read the GUI help section.
An extended explanation of the classification measures (such as error rate, non-error rate, specificity, sensitivity, etc...) has been added in the help section, here.
[-> top]