Materials used in the study are provide in dataset folder. The sensitivity of five popular classifiers which include k-Nearest Neighbor (k-NN), Multilayer perceptron (MLP), Gaussian Naive Bayes (GNB) and Support Vector Machines (SVM) with Linear and RBF kernels is measured to determine the most suitable algorithms for feature weighting.
Run Main_fw.py for finding the feature weights and parameters of classifiers simultaneously
Run Main_def.py for finding the best solution from parametric search on unweighted data.
If you use the code, please cite the following paper:
Dalwinder Singh and Birmohan Singh, "Sensitivity analysis of feature weighting for classification", Pattern Analysis and Applications, 2022.