Application of Machine Learning in the Construction of a Mortality Table

This work proposes the construction of a general mortality table using machine learning algorithms, with the objective of improving adherence to the rates observed in a Brazilian pension fund. The study applies 12 regression models available in the Scikit-learn library, with optimization through cross-validation and hyperparameter adjustment. Seeking optimization in the models, the K-Nearest Neighbors (KNN) algorithm presented the best
performance in the test set. Thus, the proposed approach can offer greater flexibility in the construction of actuarial tables, being a promising alternative to traditional methodologies.