Development of new methods for the joint modeling of longitudinal and survival data with applications in finance and insurance

点击翻译按钮获取中文摘要

A large proportion of Canadian households find it challenging to manage their finance and insurance products. As a matter of fact, according to Statistics Canada, Canadian household credit market debt as a proportion of disposable income has increased to more than 176 % during the first quarter of 2020. Making sure that Canadians receive offers of products better suited to their specific needs at more opportune times would go a long way in improving their financial health. The goal of this project is to develop analytical tools that can help a major Canadian provider of financial and insurance products to identify the best products to offer to their clients and the best times at which to offer these products using the massive historical information that is available in the company's databases. While some statistical and machine learning algorithms that already exist can be tweaked to perform similar tasks, no method that fully takes into account the temporal, hierarchical and network structure of the data are currently available. In this project, such methods will be developed and will be implemented in open software that will be made publicly available. Because historical financial databases do not reflect the diversity of the current client base of financial companies in major Canadian cities, efforts will be made to make sure that the methods to be developed will be as efficient to improve the financial health of minority groups as they will be for majority groups.