We present and describe the GPFDA package for R. The package provides flexible functionalities for dealing with Gaussian process regression (GPR) models for functional data. Multivariate functional data, functional data with multidimensional inputs, and nonseparable and/or nonstationary covariance structures can be modeled. In addition, the package fits functional regression models where the mean function depends on scalar and/or functional covariates and the covariance structure is modeled by a GPR model. In this paper, we present the versatility of GPFDA with respect to mean function and covariance function specifications and illustrate the implementation of estimation and prediction of some models through reproducible numerical examples.
翻译:我们介绍并描述R.的GPFDA软件包,该软件包为处理Gausian进程回归模型(GPR)功能数据提供了灵活的功能功能;可以建模多变量功能数据、具有多维投入的功能数据以及非分离和/或非静止共变结构;此外,软件包适合功能回归模型,其中平均功能取决于天平和/或功能共变,而共变结构则由GPR模型建模;我们在本文件中介绍了GPFDA在平均功能和共变函数规格方面的多功能性,并通过可复制的数字实例说明某些模型的估算和预测执行情况。