-
Wald inference on the nonparametric term f for space-time models
-
Spatio-Temporal Density Estimation
- Inference in space and space-time model (Gaussian family, no forcing terms): In space model inference is available for both parametric and nonparametric components; In space-time model inference only on the parametric component has been implemented.
- fdaPDE over Linear Networks domains
- smooth regression for non-gaussian space-time data (GLM model)
Optimization in spatio-temporal setting: introduced Newton method and refactored grid code.
Iterative method for spatio-temporal regression with PDE regularization (in this framework also two methods for the computation of GCV Exact and Stochastic)
- Optimization methods (Newton's methods) to find best smoothing parameter through GCV minimization
- smooth regression for non-gaussian data (GLM model)
- smooth regression for space-time data
- density estimation
- faster search algorithm
Compilation errors with clang fixed.
Compilation error in macOS fixed.
Compilation error in solaris fixed.
gcc-ASAN problem with RTriangle fixed.
Compilation errors with clang fixed.
- smooth regression for manifold domains
- smooth regression with areal data
- functional smooth PCA (SF-PCA algorithm) : function FEM.FPCA
- Stochastic GCV computation has been added: parameter 'GCVmethod' can be used both regression and FPCA, can be either 'Stochastic' or 'Exact'.
- Kfold cross-validation is available in FPCA algorithm.
- smooth.FEM.basis, smooth.FEM.PDE.basis and smooth.FEM.PDE.sv.basis are deprecated, use smooth.FEM instead.
- the R-only functions (whose names began with R_) have been deprecated, and will be removed soon.
- The usage of 'MESH' in classes and function names has been deprecated, now use 'mesh'.
- Old datasets have been removed. New datasets are provided.
- The parameter CPP_CODE has been removed.
A bug in R/C++ indexes conversion in the space-varying regression has been fixed.