Mortalitysmooth: An R package for smoothing poisson counts with P-splines

Abstract

The MortalitySmooth package provides a framework for smoothing count data in both one-and two-dimensional settings. Although general in its purposes, the package is specifically tailored to demographers, actuaries, epidemiologists, and geneticists who may be interested in using a practical tool for smoothing mortality data over ages and/or years. The total number of deaths over a specified age-and year-interval is assumed to be Poisson-distributed, and P-splines and generalized linear array models are employed as a suitable regression methodology. Extra-Poisson variation can also be accommodated. Structured in an S3 object orientation system, MortalitySmooth has two main functions which fit the data and define two classes of objects: Mort1Dsmooth and Mort2Dsmooth. The methods for these classes (print, summary, plot, predict, and residuals) are also included. These features make it easy for users to extract and manipulate the outputs. In addition, a collection of mortality data is provided. This paper provides an overview of the design, aims, and principles of MortalityS-mooth, as well as strategies for applying it and extending its use.