A Comparative Study between Kernel SVM and k-NN for the Prediction of Sleep Stages

Abstract

This research compares two supervised learning algorithms to determine the sleep stages.  Sleep is the primary activity of the brain that should be sufficiently taken to regulate the human mind. Lack of sleep may cause hypertension, depression, diabetes and many type of health problems. For this study, we take the data of 887 candidates to predict their sleep quality into categorized form. Then, we utilize both of the supervised learning algorithms for the comparison between human sleep stages by seeking their accuracies. The algorithms we used for this study are k-Nearest Neighbor (k-NN) which gave the accuracy of 81.98% and the other side the algorithm named Kernel SVM gives the better accuracy of 90.54%.

Keywords:  K-nearest neighbor, Kernel SVM, machine learning algorithm, sleep stages,  supervised algorithm, wearable device.