Gender classification via wearable gait analysis sensor

The most common activity people do in daily life is walking. Gait is an action that involves the whole body. The use of two legs in a harmonious manner is called walking for the purpose of moving from one position to another. Gait is person-specific and can be used to distinguish like fingerprints, palmprints, and iris. In this study, it is aimed to determine the gender of the persons with the gait parameters obtained by the wearable sensors. For this purpose, a total of 50 gait that consist of 23 female and 27 male gait were examined and 321 gait parameters were calculated for one gait. Gender classification success rates are calculated via three different classification algorithms as Support Vector Machine (SVM), k-nearest neighbors (k-NN) and Decision Tree classification algorithms and validated by leave one out cross validation (LOOCV) method. The most successful result in gender classification on generated walking data is provided by decision tree method.