Small issues trying to pass R CMD CHECK - still an issue with testtha…

…t and R CMD CHECK see - r-lib/testthat#86

NAMESPACE

@@ -8,6 +8,7 @@ export(check_dims)
 export(gradMongrelCollapsed)
 export(hessMongrelCollapsed)
 export(loglikMongrelCollapsed)
+export(mongrel)
 export(mongrel_sim)
 export(mongrel_tidy_samples)
 export(mongrel_to_alr)

R/RcppExports.R

@@ -141,11 +141,10 @@ hessMongrelCollapsed <- function(Y, upsilon, ThetaX, K, A, eta) {
 #' @seealso \code{\link{uncollapseMongrelCollapsed}}
 #' @examples
 #' sim <- mongrel_sim()
-#' attach(sim)
 #' 
 #' # Fit model for eta
-#' fit <- optimMongrelCollapsed(Y, upsilon, Theta%*%X, K, A, 
-#'                              random_mongrel_init(Y))  
+#' fit <- optimMongrelCollapsed(sim$Y, sim$upsilon, sim$Theta%*%sim$X, sim$K, 
+#'                              sim$A, random_mongrel_init(sim$Y))  
 optimMongrelCollapsed <- function(Y, upsilon, ThetaX, K, A, etainit, n_samples = 2000L, calcGradHess = TRUE, b1 = 0.9, b2 = 0.99, step_size = 0.003, epsilon = 10e-7, eps_f = 1e-8, eps_g = 1e-5, max_iter = 10000L, verbose = FALSE, verbose_rate = 10L, decomp_method = "eigen", eigvalthresh = 0, no_error = FALSE) {
     .Call('_mongrel_optimMongrelCollapsed', PACKAGE = 'mongrel', Y, upsilon, ThetaX, K, A, etainit, n_samples, calcGradHess, b1, b2, step_size, epsilon, eps_f, eps_g, max_iter, verbose, verbose_rate, decomp_method, eigvalthresh, no_error)
 }
@@ -196,14 +195,14 @@ optimMongrelCollapsed <- function(Y, upsilon, ThetaX, K, A, etainit, n_samples =
 #' @seealso \code{\link{optimMongrelCollapsed}}
 #' @examples
 #' sim <- mongrel_sim()
-#' attach(sim)
 #' 
 #' # Fit model for eta
-#' fit <- optimMongrelCollapsed(Y, upsilon, Theta%*%X, K, A, 
-#'                              random_mongrel_init(Y))  
+#' fit <- optimMongrelCollapsed(sim$Y, sim$upsilon, sim$Theta%*%sim$X, sim$K, 
+#'                              sim$A, random_mongrel_init(sim$Y))  
 #' 
 #' # Finally obtain samples from Lambda and Sigma
-#' fit2 <- uncollapseMongrelCollapsed(fit$Samples, X, Theta, Gamma, Xi)
+#' fit2 <- uncollapseMongrelCollapsed(fit$Samples, sim$X, sim$Theta, 
+#'                                    sim$Gamma, sim$Xi, sim$upsilon)
 uncollapseMongrelCollapsed <- function(eta, X, Theta, Gamma, Xi, upsilon, ret_mean = FALSE) {
     .Call('_mongrel_uncollapseMongrelCollapsed', PACKAGE = 'mongrel', eta, X, Theta, Gamma, Xi, upsilon, ret_mean)
 }

R/fit_mongrel.R

@@ -40,10 +40,10 @@
 #'  model if laplace approximation is not preformed. 
 #' @return an object of class mongrelfit
 #' @md
+#' @export
 #' @examples 
 #' sim <- mongrel_sim()
-#' attach(sim)
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(sim$Y, sim$X)
 #' @seealso \code{\link{mongrel_transforms}} provide convenience methods for 
 #'  transforming the representation of mongrelfit objects (e.g., conversion to 
 #'  proportions, alr, clr, or ilr coordinates.)
@@ -53,13 +53,13 @@
 #'   
 #' Generic functions including \code{\link[=summary.mongrelfit]{summary}},  
 #' \code{\link[=print.mongrelfit]{print}}, 
-#'  \code{\link[=coefficients.mongrelfit]{coeficients}},  
+#'  \code{\link[=coef.mongrelfit]{coef}},  
 #'  \code{\link[=as.list.mongrelfit]{as.list}}, and  
 #'  \code{\link[=predict.mongrelfit]{predict}}
 #' 
 #' Plotting functions provided by \code{\link{plot}} and \code{\link{ppc}} 
 #' (posterior predictive checks)
-fit_mongrel <- function(Y, X, upsilon=NULL, Theta=NULL, Gamma=NULL, Xi=NULL,
+mongrel <- function(Y, X, upsilon=NULL, Theta=NULL, Gamma=NULL, Xi=NULL,
                     init=random_mongrel_init(Y), 
                     pars=c("Eta", "Lambda", "Sigma"),
                     ...){

R/mongrel.R

@@ -1,3 +1,13 @@
+#' mongrel: Fitting and Analysis of Multinomial Logistic Normal Regression Models
+#' 
+#'  Provides methods for fitting and inspection of Bayesian Multinomial 
+#'  Logistic Normal Regression Models ("Mongrel Models") using MAP estimation 
+#'  (with the ADAM optimizer) and Laplace Approximation. Key functionality is 
+#'  implemented in C++ for scalability. 
+#'  
+#' @docType package
+#' @name mongrel_package
+#' 
 #' @useDynLib mongrel
 #' @importFrom Rcpp sourceCpp
 NULL

R/mongrel_methods.R

@@ -14,8 +14,7 @@
 #' @return tibble
 #' @examples 
 #' sim <- mongrel_sim()
-#' attach(sim)
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(sim$Y, sim$X)
 #' fit_tidy <- mongrel_tidy_samples(fit, use_names=TRUE)
 #' head(fit_tidy)
 mongrel_tidy_samples<- function(m, use_names=FALSE){
@@ -62,7 +61,7 @@ summary_check_precomputed <- function(m, pars){
 #' @param m an object of class mongrelfit 
 #' @param pars character vector (default: c("Eta", "Lambda", "Sigma"))
 #' @param use_names should summary replace dimension indicies with mongrelfit 
-#'   names if names Y and X were named in call to \code{\link{fit_mongrel}}
+#'   names if names Y and X were named in call to \code{\link{mongrel}}
 #' @param gather_prob if TRUE then prints quantiles in long format rather than 
 #'  wide (useful for some plotting functions)
 #' @param ... other expressions to pass to summarise (using name 'val' unquoted is 
@@ -74,7 +73,7 @@ summary_check_precomputed <- function(m, pars){
 #' @importFrom dplyr group_by select ungroup
 #' @examples 
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' summary(fit, pars="Eta", median = median(val))
 #' 
 #' # Some later functions make use of precomputation
@@ -120,7 +119,7 @@ summary.mongrelfit <- function(m, pars=NULL, use_names=TRUE, gather_prob=FALSE,
 #' @export
 #' @examples 
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' print(fit)
 #' }
 #' @seealso \code{\link{summary.mongrelfit}} summarizes posterior intervals 
@@ -154,14 +153,14 @@ print.mongrelfit <- function(m){
 #' 
 #' @param m an object of class mongrelfit
 #' @param use_names if column and row names were passed for Y and X in 
-#' call to \code{\link{fit_mongrel}}, should these names be applied to output 
+#' call to \code{\link{mongrel}}, should these names be applied to output 
 #' array. 
 #' @return Array of dimension (D-1) x Q x iter
 #' 
 #' @export
 #' @examples 
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' coef(fit)
 #' }
 coef.mongrelfit <- function(m, use_names=TRUE){
@@ -178,7 +177,7 @@ coef.mongrelfit <- function(m, use_names=TRUE){
 #' @export
 #' @examples 
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' as.list(fit)
 #' }
 as.list.mongrelfit <- function(m){
@@ -209,8 +208,7 @@ as.list.mongrelfit <- function(m){
 #' @importFrom stats median predict runif
 #' @examples 
 #' sim <- mongrel_sim()
-#' attach(sim)
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(sim$Y, sim$X)
 #' predict(fit)
 predict.mongrelfit <- function(m, newdata=NULL, response="LambdaX", size=NULL, 
                                use_names=TRUE, summary=FALSE, ...){
@@ -309,7 +307,7 @@ predict.mongrelfit <- function(m, newdata=NULL, response="LambdaX", size=NULL,
 #' @name access_dims
 #' @examples 
 #' \dontrun{
-#' m <- fit_mongrel(Y, X)
+#' m <- mongrel(Y, X)
 #' ncategories(m)
 #' nsamples(m)
 #' ncovariates(m)

R/mongrel_plotting.R

@@ -15,7 +15,7 @@
 #' @importFrom dplyr filter
 #' @examples
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' plot(fit, par="Lambda")
 #' plot(fit, par="Sigma")
 #' plot(fit, par="Sigma", focus.coord=c("s1", "s2", "s3"))
@@ -101,7 +101,7 @@ plot_mf_sigma <- function(m, focus.coord=NULL, use_names=TRUE){
 #' @importFrom driver gather_array
 #' @examples 
 #' \dontrun{
-#' fit <- fit_mongrel(Y, X)
+#' fit <- mongrel(Y, X)
 #' ppc(fit)
 #' }
 ppc <- function(m, iter=50){

R/mongrel_transforms.R

@@ -19,7 +19,7 @@
 #' @import driver 
 #' @examples
 #' \dontrun{
-#' m <- fit_mongrel(Y, X)
+#' m <- mongrel(Y, X)
 #' m.prop <- mongrel_to_proportions(m)
 #' # convert back to default coordinates (alr with D-th part as reference)
 #' m <- mongrel_to_alr(m.prop, ncategories(m))