Three directions in model-based clustering, classification and discriminant analysis

点击翻译按钮获取中文摘要

Contemporary science research is confronted with ever-growing masses of data. Given that scientists rely on statistics to interpret results, the availability of suitable statistical methods is essential. Unfortunately, some research questions cannot currently be answered due to the lack of appropriate statistical techniques. The proposed research will explore three directions in statistical research to help better answer some of these questions. Specifically, the techniques developed will aid in the identification of groups within data, or 'clustering'. Advanced statistical clustering techniques are important in a variety of areas of scientific endeavour. The proposed research will facilitate, directly or indirectly, work in areas like: bioinformatics; finance; food authenticity; social science, especially in the construction of social networks; political science, specifically in the identification of voting blocks; and sensory science, where the applicant is currently conducting work on clustering categorical taste testing data to better segment the sensory space. For example, the proposed research will help scientists discover the functions of genes through the development of techniques for clustering genes based on their behaviour over time.