Modeling and Exploration of Self Rated Health Data using Machine Learning
|Title||Modeling and Exploration of Self Rated Health Data using Machine Learning|
|Summary||An analysis of self rated health data|
|Keywords||Data mining, Machine Learning, Data Exploration|
|Prerequisites||Artificial Intelligence, Learning System courses|
|Supervisor||Sławomir Nowaczyk, Anita Sant'Anna, Parivash Pirasteh|
he data set used for this thesis is taken from an online stress management system called HealthWatch.se. HealthWatch.se aims to provide tools for individuals and organizations to increase health and quality of life, as well as reduce stress-related problems. Healthwatch consists of eleven questions about one's health. These questions are used to measure multiple aspects of health quality including energy, stress, sleep and etc. The data study uses both weekly and monthly data collected from 6854 users form 2006 to 2013. This thesis project aims to perform initial data exploration, identify several promising directions for further analysis, prioritise them based on domain relevance & data mining value, and do a comprehensive analysis on at least two directions. Examples of directions to consider include prediction of self rated health based on other features (over time and across individuals/groups); estimating the risk of long term stress-related work absence and proposing individualized ways to prevent it; understanding different patterns of health behaviors, both common and unique/atypical ones, as well as identifying relevant factors; trend discovery, changes over time, individual and common patterns; finding cut-off values in health indicator variables; feature selection and pattern mining using machine learning methods.