

This study proposes a combinatory feature selection method for HRV and ECG signals classification to detect the level of stress. The main purpose of feature selection methods is to reduce the dimension of features to those data that their changes are sensitive to the algorithm that is being used. This will make the performance of classifiers more efficient with higher accuracy and will reduce the time of processing. The paper studies the effect of three feature selection methods ANOVA, SFS and SBS on the complete list of ECG and HRV features derived from 16 subjects. For every set of features resulted from different feature selection, the level of error for stress detection is investigated using ANFIS classifier. Finally the method of combinatory feature selection method (intersection area of mentioned selection method) is compared in the matter of error index. The results reveal that the combinatory method consists of the intersection of ANOVA, SFS and SBS algorithms has the benefit of good and accurate detection results; fast and low memory space, low time complexity, and the computations are not so complicated.