RcppCensSpatial

Spatial Estimation and Prediction for Censored/Missing Responses

It provides functions for estimating parameters in linear spatial models with censored or missing responses using the Expectation-Maximization (EM), Stochastic Approximation EM (SAEM), and Monte Carlo EM (MCEM) algorithms. These methods are widely used to obtain maximum likelihood (ML) estimates in the presence of incomplete data. The EM algorithm computes ML estimates when a closed-form expression for the conditional expectation of the complete-data log-likelihood is available. The MCEM algorithm replaces this expectation with a Monte Carlo approximation based on independent simulations of the missing data. In contrast, the SAEM algorithm decomposes the E-step into simulation and stochastic approximation steps, improving computational efficiency in complex settings. In addition, the package provides standard error estimation based on the Louis method. It also includes functionality for spatial prediction at new locations. References used for this package: Galarza, C. E., Matos, L. A., Castro, L. M., & Lachos, V. H. (2022). Moments of the doubly truncated selection elliptical distributions with emphasis on the unified multivariate skew-t distribution. Journal of Multivariate Analysis, 189, 104944 <doi:10.1016/j.jmva.2021.104944>; Valeriano, K. A., Galarza, C. E., & Matos, L. A. (2023). Moments and random number generation for the truncated elliptical family of distributions. Statistics and Computing, 33(1), 32 <doi:10.1007/s11222-022-10200-4>.

Versions across snapshots

Dependencies (latest)