Browsing by Subject "ryhmittely"

Objective. The topic of this study is classification in psychology and the aim is to introduce, why classification is important in psychology, what classification is used for in psychology and how classification is done in psychology. In addition the study introduces attributes and assumptions of the most common statistical methods used in psychology. The aim was to create a coherent totality, which can be used when either choosing a method for study or evaluating methods of previous studies. The study also describes common challenges regarding classification and ways to evaluate classification solutions. The statistical methods focus on unsupervised learning methods, which are used when the structure and number of classes are unknown. Hierarchical, centroid-based and distribution based clustering are introduced alongside with latent class analysis. For supervised learning methods, neural networks, Bayesian methods and decision trees are considered. Methods. Since this study is based on statistics, most of the literature used is considering statistical methods. In addition, studies of psychology which use methods relevant for this study are used as reference. Results. Classification in psychology is important in measuring, communicating about and understanding phenomena. It has made diagnostics, recognizing risk groups and specifying attributes and features possible. Classification has also challenges. Psychological phenomena can be heterogenetic, unstable and used parameters do not necessarily measure intended phenomenon. Classification methods are numerous and every method has its own attributes, weaknesses and strengths. And every solution is only a data dependent model. Statistical methods are in totality useful tools in psychological studies and literature. Understanding of these statistical methods enables producing accurate models and also evaluating them.