These Matlab scripts are intended to be used with LIBSVM and the Excel dataset available from https://archive.ics.uci.edu/ml/datasets/Mice+Protein+Expression. Before use, it is assumed that the dataset has been modified so that all missing values have been replaced with -1 and the names of classes have been replaced with the numeric values found in the scripts. The usage of each file is described here:
-
dataProcessing.m - Performs all standard preprocessing procedures. Also generates training and testing files for 12 pairwise comparisons in sparse format. These files can be used for training and prediction using the svm_train and svm_predict commands in LIBSVM.
-
modelFileProcessing.m - Converts the model files generated by LIBSVM into a readable format. The output consists of the weight values of all features in the dataset followed by a single bias value in csv format.
-
dataProcessing_filtered.m - Performs the same operations as dataProcessing.m, but can be modified to include only specific features according to their numeric value in the dataset.
-
wilcoxon2.m - Executes the Wilcoxon Rank-Sum test on all features in the dataset for all 12 comparisons considered.