Sleep Prediction by Various Supervised Machine Learning Model

Abstract

In this research article, we analyze the behavioral and physiological dataset of 887 contestants and categorize their sleep quality into Best, Average and Poor. Sleep is one of the brain's main activities that play a requisite role in the learning and physical movement of a person. Sleep is a condition where eyes are closed and centers of the nervous system are disabled. Sleep thus makes the person partly or fully sleepy and helps the brain to have a less complicated network. One-third of human life is spent on sleep and disorders such as Obstructive Sleep Apnea (OSA) and insomnia are common and may impact physical health badly. Sleep disorders may lead to sleepiness, anxiety, depression or may cause death.
For these purposes, the development of a methodology is necessary which can identify and evaluate sleep patternsin order to recognize the sleep stages of earthborn. Here, we are using Machine Learning Classification Techniques to predict sleep stages like Support Vector Machine (SVM), Random Forest Classification, eXtreme Gradient Boosting (XGBoost), Naïve Bayes. The accuracies of these classifiers are 98.19%, 53.44%, 100.00%, 89.63% respectively. In this maximum accuracy is given by XGBoost of 100% which is very rare. So, it is consider as more relevant method for measuring the sleep quality.

Keywords: Sleep quality, Wearable Devices, XGBoost, Random Forest Classification, Navie Bayes, Supervised machine learning.