Análise do tempo até o cancelamento de seguros de vida: uma abordagem via análise de sobrevivência

Abstract

Companies that sell long-term products or subscriptions need to assess how long their customers retain their products or subscriptions. Usually, the cost of acquiring new customers is extremely high, making it crucial for companies that these customers maintain their products. This study aims to apply survival models to describe the random variable "time until cancellation." Data related to life insurance contracts that do not exhibit proportional hazards are analyzed, thus requiring tools capable of capturing this characteristic. Alternatives such as piecewise exponential models and Cox models with time-varying effects are considered. This work consistently seeks to utilize tools already available in the survival analysis literature to solve the problem, explaining the necessary theory for modeling in a simplified manner, so that it can be applied in the daily corporate world.