New features in XLSTAT 2021
New features added to XLSTAT in 2021 include:
- One-class Support Vector Machine (available in all XLSTAT solutions except Basic) The One-class Support Vector Machine (One-class SVM) algorithm seeks to envelop underlying inliers. The aim is to separate data into two classes (based on a decision function): the positive one, considered as the class of inliers, and the negative one, considered as the class of outliers. Commonly used for fraud detection and machine fault diagnosis. Access this feature under the Machine Learning menu.
- Support Vector Machine (available in all XLSTAT solutions except Basic) A new algorithm is integrated for SVM classification (Fan 2005). It allows you to solve quadratic problems faster thanks to second order information. A k-fold cross-validation is added for all proposed methods (regression, binary and multi-class classification). The ROC curve is now displayed after the confusion matrix in the case of classification. Access this feature under the Machine Learning menu.
- Response Surface Designs (available in XLSTAT Life Sciences, Quality & Premium) Response surface designs are widely used to optimize various processes, like in the food industry, for example. A revamped interface will allow you to easily generate a design for response surface analysis. A new table has been added to the report sheet allowing you to define the optimization settings of the analysis or use the default software settings. Interactions between factors can be now taken into account to the model. Access this feature under the DOE menu.
- Discriminant Analysis (available in all XLSTAT solutions) The Outputs tab of the dialog box is now divided into two sub-tabs: General and Tests. The first one offers generic outputs (descriptive statistics, eigenvalues, confusion matrix, ect.), while the second one offers various tests to validate specific assumptions of the model (Box test, Wilk's test, Pillai's trace, etc). The results sheet has been restructured similarly to the Outputs tab. It is also possible to import large-volume data files via the dedicated button in the dialog box (check our example). Access this feature under the Analyzing Data menu.
- Principal Component Regression (available in all XLSTAT solutions except Basic) Principal Component Regression (PCR) is built on Principal Components Analysis (PCA). A major advantage of PCR over classical regression is the ability to generate charts that clearly describe the data structure, such as the correlation chart and the biplot. Several improvements and fixes have been made in this existing feature:
- The Cooks checkbox has been removed from the Outputs tab of the dialog box. This has been replaced by "Regression diagnostics'' which contains Cooks and other indices.
- Suppression of Press in the Outputs tab. It is now automatically displayed in the Goodness of Fit table.
- The "Inner Circle" option is now greyed out (only available for PLS).
- It is now possible to import large-volume data files via the dedicated button of the dialog box (check our example).
- A computational bug has been fixed in the case of having more than one Y variable.
Access this feature under the Modeling data menu.
The changes from version 2019 to 2020, and those added in 2020, can be found here.