Análise de cluster em um plano de saúde via wavelets

• Ano de defesa: 2013
• Autor(a) principal: Paola Mara de Oliveira Quinto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Wavelets; Análise de cluster; Método k-means; Análise de resolução múltiplas; Perfis de clientes; Estatística
• Link de acesso: http://hdl.handle.net/1843/BUOS-974H7J

One of the issues that has led to concerns by operators of health plans, and at the same time, has been the subject of many studies, is rising costs and their concentration in a small portion of the portfolio. Within this context, different types of customers are responsible for writing plans and generate costs. However, no one knows how many there are and what the characteristics of each one of them, and our goal in this work is to identify them. The database is used for a particular health plan, and the method used to separate customers into groups is called cluster analysis. The purpose of cluster analysis is to seek a classification according to the natural features that the sample, forming groups of objects by similarity. However, when applied to the database in question, the method fails to separate customers in groups with homogeneous characteristics according costs. Thean, we look a way to rewrite the costs through the wavelet coefficients, which summarize all the information contained in the time series of the costs of each client's health plan. Several analysis were performed, but we will bring the better result. We describe the customer profiles formed, as well as their characteristics with respect to the series of costs and descriptive general group, such as age, sex, total cost ownership, among others.