Variable Reduction Testbench is a MATLAB module that allows the application of several methods for variable reduction based on correlation analysis. These methods rank the variables, starting from the most correlated ones (i.e. the most redundant variables, thus they can be easily removed from the dataset without significant information loss) and ending with the variables with lowest correlation. Most of the methods suggest also how many variables could be removed from the dataset without information loss. Datasets are required in the form of matrices of n x p elements, where n are the samples (rows) and p the variables (columns). Each method also provides the value of standardized Shannon entropy (H) for all the variables, since it is a very useful index of the quantity of information contained in each variable.
The module is entirely based on a graphic interface; the ranking output is given in several forms (plain text file, MATLAB file and a plot) thus allowing an easy understanding of the results. Furthermore, outputs of the correlation analysis are additionally provided.
Download
Fill in the following form. Your personal data will be used only for notification via email of new releases of the toolbox and will not be communicated to external third parties. Once the form has been submitted, download and open the rar file and extract all the Matlab modules in a unique folder. Additionally, you can find a pdf help file with all the instructions for the use of Variable Reduction Testbench some data files.