Predicting patient waiting times with unsupervised machine learning

Data has become a fundamental element of political, economic, social, scientific, and technological progress in today's world. With the acceleration of digitalization, vast amounts of data are being generated in almost every field, including healthcare, education, engineering, finance, and public administration. Once this data is analyzed and meaningful insights are extracted, it becomes an indispensable component of decision-making processes. To transform raw data into meaningful information, it must be processed using statistical methods and machine learning algorithms. This thesis examines concepts and methodologies such as data, big data, data mining, machine learning, types of machine learning, and concepts such as health, healthcare services, and decision-making in healthcare. Healthcare systems were chosen as the research framework, and the application of big data and machine learning techniques in healthcare was examined. To address the use of machine learning methods in healthcare, the waiting times of patients presenting to a hospital emergency department were analyzed using the mixture distribution method, an unsupervised learning method in machine learning. Cluster analysis was conducted on patient waiting times across different branches using the Gamma mixture distribution and Poisson mixture distribution models. It has been observed that the statistical distributions of patient waiting times vary by branch and that these data are more appropriately represented by mixture statistical distributions rather than a single statistical distribution. Keywords: Data, Big Data, Mixture Distribution, Clustering, Machine Learning