Changes from ver 1.1-0 to ver 1.1-1 [May-25-2018]
1) bigsplines-package
* BUG FIX IN bigssp (interactions between parametric and nonparametric effects)
* BUG FIX IN ssBasis (periodic = TRUE input now works)
* Fixed NOTE on CRAN Results page (about registering bigsplines.so)
* Package file is now update-to-date with Description file
* Updated references throughout the package.
2) bigssa and bigssp
* Fixed warning for models with a single random effect.
3) imagebar
* Improved the default color scheme (for better distinction).
* Function can now take in a single input (x or z)
4) summary.bigssa, summary.bigssg, and summary.bigssp
* Added the "diagnostics" option (default diagnostics = FALSE)
* If diagnostics = TRUE, cosine diagnostics are returned.
Changes from ver 1.0-9 to ver 1.1-0 [Feb-01-2017]
1) bigsplines-package
* Added linear smoothing spline (type="lin") to bigssa, bigssg, and bigssp
* Added ordspline function for fitting ordinal smoothing spline
* Added ssBasis function for generating polynomial smoothing spline basis matrices
* Added more options to plotci function with bars=TRUE option
* Update references throughout
2) bigssa and makessa
* Added new reproducing kernel: linear smoothing spline.
* Use type="lin" for linear smoothing spline.
3) bigssg and makessg
* Added new reproducing kernel: linear smoothing spline.
* Use type="lin" for linear smoothing spline.
4) bigssp and makessp
* Added new reproducing kernel: linear smoothing spline.
* Use type="lin" for linear smoothing spline.
5) ordspline
* New function for fitting ordinal smoothing spline.
* Use monotone=TRUE to fit monotonic ordinal smoothing spline.
6) plotci
* When bars=TRUE, now possible to add a unique barlty, barlwd,
and col.ci for each of the input values of x
7) ssBasis
* New function for generating polynomial smoothing spline basis matrices.
* Use m=1 for linear, m=2 for cubic, and m=3 for quintic.
* First and second derivative of basis matrices also available.
Changes from ver 1.0-8 to ver 1.0-9 [Aug-11-2016]
1) bigsplines-package
* Added plotbar function for generic X-Y plotting with colorbars
* Added plotci function for generic X-Y plotting with confidence intervals
* Improvements to imagebar function (added zaxis.at and zaxis.labels inputs)
* Improvements to prediction function for bigssa, bigssg, and bigssp
* Improvements to ordinal smoothing spline computation
2) bigssa and makessa
* Computational improvements for ordinal smoothing splines
* Improvements to prediction functions
3) bigssp and makessp
* Computational improvements for ordinal smoothing splines
* Improvements to prediction functions
4) imagebar
* New input zaxis.at to controls tick locations for colorbar
* New input zaxis.labels to control tick labels for colorbar
5) plotbar
* New function for generic X-Y plotting with colorbars
* Plots X vs Y with points color-coded according to Z
6) plotci
* New function for generic X-Y plotting with confidence intervals
* Plots X vs Y with confidence bands or bars
* Allows for a user-defined link function
Changes from ver 1.0-7 to ver 1.0-8 [Mar-17-2016]
1) bigsplines-package
* Added functionality for ordinal predictors (type="ord")
* More options for bigspline function (linear splines)
* More options for estimating variance components
* Improvements to imagebar function
* Updated some references throughout
* Bug fix for cub0 kernel predictions (when x not in [0,1])
2) bigspline
* Added type="lin"" option for linear smoothing spline
3) bigssa and makessa
* Added type="ord" option for ordinal predictors
* Added Newton-Raphson option for variance component estimation
* Bug fix for cub0 kernel predictions (when x not in [0,1])
* Changes/improvements to Example #3 in bigssa help file
4) bigssg and makessg
* Added type="ord" option for ordinal predictors
* Bug fix for cub0 kernel predictions (when x not in [0,1])
5) bigssp and makessp
* Added type="ord" option for ordinal predictors
* Added Newton-Raphson option for variance component estimation
* Bug fix for cub0 kernel predictions (when x not in [0,1])
6) imagebar
* Changes to internals when drawbar=TRUE
* Now possible to add points/lines to image
Changes from ver 1.0-6 to ver 1.0-7 [Aug-09-2015]
1) bigsplines-package
* Many more options for random effects
* Improvements to REML algorithms
* Changes in default REML options and syntax
* Updated some references throughout
2) bigssa and makessa
* Many more options for random effects
* Changes in default REML options and syntax
3) bigssp and makessp
* Many more options for random effects
* Changes in default REML options and syntax
Changes from ver 1.0-5 to ver 1.0-6 [Mar-16-2015]
1) bigsplines-package
* Improvements to imagebar (more customization options)
* Improvements to all prediction functions (more options)
* Minor improvements to internals of all functions
2) imagebar
* New input 'zline' to control z-label placement
* New input 'pltimage' to control image plt (see par)
* New input 'pltbar' to control colorbar plt (see par)
* Reduced default number of colors to 21 (instead of 100)
* Slight modification to default colors
3) predict.bigspline
* New input 'design' to return design matrix
* New input 'smoothMatrix' to return smoothing matrix
4) predict.bigssa
* New input 'design' to return design matrix
* New input 'smoothMatrix' to return smoothing matrix
5) predict.bigssg
* New input 'design' to return design matrix
* New input 'smoothMatrix' to return smoothing matrix
6) predict.bigssp
* New input 'design' to return design matrix
* New input 'smoothMatrix' to return smoothing matrix
7) predict.bigtps
* New input 'design' to return design matrix
* New input 'smoothMatrix' to return smoothing matrix
Changes from ver 1.0-4 to ver 1.0-5 [Feb-11-2015]
1) bigsplines-package
* Minor changes to bigspline, bigtps, and imagebar.
* Major changes to bigssa/makessa and bigssp/makessp (added random effects).
* Major changes/improvements to bigssg function.
* Updated some references throughout.
2) bigspline (and corresponding predict, print, and summary functions)
* Changed class of output object (from class 'css' to class 'bigspline')
3) bigssa (and corresponding predict, print, and summary functions)
* For unidimensional predictor, assumes type='cub' as default.
* Changed class of output object (from class 'ssa' to class 'bigssa')
* Added support for simple random effects/intercepts.
4) bigssp (and corresponding predict, print, and summary functions)
* For unidimensional predictor, assumes type='cub' as default.
* Changed class of output object (from class 'ssp' to class 'bigssp')
* Added support for simple random effects/intercepts.
5) bigtps (and corresponding predict, print, and summary functions)
* Changed class of output object (from class 'tps' to class 'bigtps')
6) imagebar
* Restores original par on exit (so subsequent plots appear correctly).
7) bigssg (and corresponding predict, print, and summary functions)
* For unidimensional predictor, assumes type='cub' as default.
* Changed class of output objects (from class 'ssg' to class 'bigssg')
* More options for smoothing parameter selection (see input gcvtype).
* Changed default smoothing parameter selection method to "acv".
* Bug fix when using skip.iter=FALSE (early termination of iterative update).
* Bug fix for family="negbin" with unknown dispersion (multi-way SSA only).
* R-squared is now reasonably defined: R^2 = 1-SSE/SST.
* Stability and usability improvements throughout.
Changes from ver 1.0-3 to ver 1.0-4 [Oct-03-2014]
1) bigsplines-package
* Minor updates for prediction functions.
* Updated some references throughout.
2) predict.ssa
* Added option for including intercept.
* Stability and usability improvements.
3) predict.ssg
* Added option for including intercept.
* Stability and usability improvements.
4) predict.ssp
* Added option for including intercept.
* Stability and usability improvements.
Changes from ver 1.0-2 to ver 1.0-3 [Aug-26-2014]
1) bigsplines-package
* Major update: support for non-Gaussian responses (see bigssg and makessg)
* Updated contact information (helwig@umn.edu)
2) bigssa
* Added weights input (for weighted penalized LS)
* Added check for input gcvopts (to ensure options are valid)
3) bigssg
* Extension of the bigssa function for non-Gaussian data
* Handles data from five distributions: Binomial, Poisson, Gamma,
Inverse Gaussian, and Negative Binomial
4) bigssp
* Added weights input (for weighted penalized LS)
* Added check for input gcvopts (to ensure options are valid)
5) makessa
* Added weights input (for weighted penalized LS)
* Added check for input gcvopts (to ensure options are valid)
6) makessg
* Extension of the makessa function for non-Gaussian data
* Handles data from five distributions: Binomial, Poisson, Gamma,
Inverse Gaussian, and Negative Binomial
7) makessp
* Added weights input (for weighted penalized LS)
* Added check for input gcvopts (to ensure options are valid)
Changes from ver 1.0-1 to ver 1.0-2 [July-12-2014]
1) bigsplines-package
* New title: bigsplines: Smoothing splines for large samples
* Updated references
2) bigspline
* binsamp is called internally, so all binsamp changes apply
* Accuracy, stability, and speed improvements for rounding parameters
* Added knotcheck input (for stability)
3) bigssa
* Accuracy, stability, and speed improvements for rounding parameters
* Added knotcheck input (for stability)
* Added gammas input (for smoothing parameter initialization)
4) bigssp
* Accuracy, stability, and speed improvements for rounding parameters
* Added knotcheck input (for stability)
* Added thetas input (for smoothing parameter initialization)
5) bigtps
* Accuracy, stability, and speed improvements for rounding parameters
* Added knotcheck input (for stability)
6) binsamp
* Bug fix: with exactly 1 data observation in each bin
* Changed default to nmbin=11
* Accuracy, stability, and speed improvements
* Added option for algorithm in ver 1.0-0 (for reproducibility of JCGS paper)
* Added return (to print knot indices)
7) imagebar
* Bug fix: last color of bar is now plotted appropriately
8) makessa
* Accuracy, stability, and speed improvements for rounding parameters
* Improvement: handles misordered interactions (e.g., time:space vs space:time)
* Added knotcheck input (for stability)
* Added gammas input (for smoothing parameter initialization)
9) makessp
* Accuracy, stability, and speed improvements for rounding parameters
* Improvement: handles misordered interactions (e.g., time:space vs space:time)
* Added knotcheck input (for stability)
* Added thetas input (for smoothing parameter initialization)
Changes from ver 1.0-0 to ver 1.0-1 [May-17-2014]
1) bigspline
* Bug fix: confirmatory smoothing with length(lambda)=1
* Bug fix: bin-sampling of knots
* binsamp is called internally, so all binsamp changes apply
2) binsamp
* Improved bin-sampling algorithm
3) bigtps
* Bug fix: confirmatory smoothing with length(lambda)=1