Merge pull request #11 from tidymodels/recipes-update

Recipes update

Author: topepo

.travis.yml

@@ -69,6 +69,10 @@ before_install:
   - sudo apt-get -y install libnlopt-dev 
   - sudo apt-get update
   - sudo apt-get -y install python3
+  - mkdir -p ~/.R
+  - echo "CXX14FLAGS=-O3 -mtune=native -march=native -Wno-unused-variable -Wno-unused-function  -Wno-macro-redefined" >> ~/.R/Makevars
+  - echo "CXX14=g++ -std=c++1y -fPIC" >> ~/.R/Makevars
 
 after_success:
   - Rscript -e 'covr::codecov()'
+  - travis_wait 59 Rscript -e 'covr::coveralls(line_exclusions="src/init.cpp")'

DESCRIPTION

@@ -7,7 +7,7 @@ Authors@R: c(
     person("RStudio", role = "cph"))
 Depends: 
     R (>= 3.1),
-    recipes (>= 0.1.3)
+    recipes (>= 0.1.3.9002)
 Imports: 
     rstanarm,
     keras,
@@ -26,6 +26,8 @@ Suggests:
     knitr,
     rmarkdown,
     covr
+Remotes:
+    tidymodels/recipes
 License: GPL-2
 Encoding: UTF-8
 LazyData: true
@@ -34,6 +36,3 @@ Roxygen: list(markdown = TRUE)
 ByteCompile: true
 URL: https://tidymodels.github.io/embed
 BugReports: https://github.com/tidymodels/embed/issues
-Remotes:
-    r-lib/generics,
-    tidymodels/recipes

NAMESPACE

@@ -22,6 +22,10 @@ export(step_lencode_bayes)
 export(step_lencode_glm)
 export(step_lencode_mixed)
 export(tidy)
+export(tidy.step_embed)
+export(tidy.step_lencode_bayes)
+export(tidy.step_lencode_glm)
+export(tidy.step_lencode_mixed)
 import(rlang)
 importFrom(dplyr,"%>%")
 importFrom(dplyr,arrange)
@@ -57,6 +61,7 @@ importFrom(recipes,is_trained)
 importFrom(recipes,names0)
 importFrom(recipes,prep)
 importFrom(recipes,printer)
+importFrom(recipes,rand_id)
 importFrom(recipes,sel2char)
 importFrom(recipes,step)
 importFrom(recipes,terms_select)

NEWS.md

@@ -1,3 +1,14 @@
+# `embed` 0.0.2
+
+A mostly maintainence release to be compatible with version 0.1.3 of `recipes`. 
+
+## Other Changes:
+
+ * The package now depends on the `generics` pacakge to get the `broom` `tidy` methods. 
+
+ * Karim Lahrichi added the ability to use callbacks when fitting tensorflow models. [PR](https://github.com/tidymodels/embed/pull/9)
+
+
 # `embed` 0.0.1
 
 First CRAN version

R/bayes.R

@@ -30,6 +30,7 @@
 #'  the computations for subsequent operations
 #' @param trained A logical to indicate if the quantities for
 #'  preprocessing have been estimated.
+#' @param id A character string that is unique to this step to identify it.
 #' @return An updated version of `recipe` with the new step added
 #'  to the sequence of existing steps (if any). For the `tidy`
 #'  method, a tibble with columns `terms` (the selectors or
@@ -93,7 +94,7 @@
 #' 
 #' # See https://tidymodels.github.io/embed/ for examples
 
-#' @importFrom recipes add_step step terms_select sel2char ellipse_check
+#' @importFrom recipes add_step step terms_select sel2char ellipse_check rand_id
 step_lencode_bayes <-
   function(recipe,
            ...,
@@ -103,7 +104,8 @@ step_lencode_bayes <-
            options = list(seed = sample.int(10^5, 1)),
            verbose = FALSE,
            mapping = NULL,
-           skip = FALSE) {
+           skip = FALSE,
+           id = rand_id("lencode_bayes")) {
     if (is.null(outcome))
       stop("Please list a variable in `outcome`", call. = FALSE)
     add_step(
@@ -116,20 +118,14 @@ step_lencode_bayes <-
         options = options,
         verbose = verbose,
         mapping = mapping,
-        skip = skip
+        skip = skip,
+        id = id
       )
     )
   }
 
 step_lencode_bayes_new <-
-  function(terms = NULL,
-           role = NA,
-           trained = FALSE,
-           outcome = NULL,
-           options = NULL,
-           verbose = FALSE,
-           mapping = NULL,
-           skip = FALSE) {
+  function(terms, role, trained, outcome, options, verbose, mapping, skip, id) {
     step(
       subclass = "lencode_bayes",
       terms = terms,
@@ -139,7 +135,8 @@ step_lencode_bayes_new <-
       options = options,
       verbose = verbose,
       mapping = mapping,
-      skip = skip
+      skip = skip,
+      id = id
     )
   }
 
@@ -152,9 +149,17 @@ prep.step_lencode_bayes <- function(x, training, info = NULL, ...) {
   res <-
     map(training[, col_names], stan_coefs, y = training[, y_name],
         x$options, x$verbose)
-  x$mapping <- res
-  x$trained <- TRUE
-  x
+  step_lencode_bayes_new(
+    terms = x$terms,
+    role = x$role,
+    trained = TRUE,
+    outcome = x$outcome,
+    options = x$options,
+    verbose = x$verbose,
+    mapping = res,
+    skip = x$skip,
+    id = x$id
+  )
 }
 
 #' @importFrom stats as.formula glm binomial coef gaussian na.omit
@@ -241,11 +246,11 @@ map_glm_coef <- function(dat, mapping) {
 #' @importFrom recipes bake prep
 #' @importFrom purrr map
 #' @export
-bake.step_lencode_bayes <- function(object, newdata, ...) {
+bake.step_lencode_bayes <- function(object, new_data, ...) {
   for (col in names(object$mapping))
-    newdata[, col] <- map_glm_coef(newdata[, col], object$mapping[[col]])
+    new_data[, col] <- map_glm_coef(new_data[, col], object$mapping[[col]])
 
-  newdata
+  new_data
 }
 
 #' @importFrom recipes printer
@@ -262,6 +267,7 @@ print.step_lencode_bayes <-
 #' @rdname step_lencode_bayes
 #' @param x A `step_lencode_bayes` object.
 #' @export
+#' @export tidy.step_lencode_bayes
 tidy.step_lencode_bayes <- function(x, ...) {
   if (is_trained(x)) {
     for(i in seq_along(x$mapping))
@@ -277,6 +283,7 @@ tidy.step_lencode_bayes <- function(x, ...) {
       terms = term_names
     )
   }
+  res$id <- x$id
   res
 }
 

R/glm.R

@@ -26,6 +26,7 @@
 #'  the computations for subsequent operations
 #' @param trained A logical to indicate if the quantities for
 #'  preprocessing have been estimated.
+#' @param id A character string that is unique to this step to identify it.
 #' @return An updated version of `recipe` with the new step added
 #'  to the sequence of existing steps (if any). For the `tidy`
 #'  method, a tibble with columns `terms` (the selectors or
@@ -74,7 +75,8 @@ step_lencode_glm <-
            trained = FALSE,
            outcome = NULL,
            mapping = NULL,
-           skip = FALSE) {
+           skip = FALSE,
+           id = rand_id("lencode_bayes")) {
     if (is.null(outcome))
       stop("Please list a variable in `outcome`", call. = FALSE)
     add_step(
@@ -85,26 +87,23 @@ step_lencode_glm <-
         trained = trained,
         outcome = outcome,
         mapping = mapping,
-        skip = skip
+        skip = skip,
+        id = id
       )
     )
   }
 
 step_lencode_glm_new <-
-  function(terms = NULL,
-           role = NA,
-           trained = FALSE,
-           outcome = NULL,
-           mapping = NULL,
-           skip = FALSE) {
+  function(terms, role, trained, outcome, mapping, skip, id) {
     step(
       subclass = "lencode_glm",
       terms = terms,
       role = role,
       trained = trained,
       outcome = outcome,
       mapping = mapping,
-      skip = skip
+      skip = skip,
+      id = id
     )
   }
 
@@ -115,9 +114,14 @@ prep.step_lencode_glm <- function(x, training, info = NULL, ...) {
   check_type(training[, col_names], quant = FALSE)
   y_name <- terms_select(x$outcome, info = info)
   res <- map(training[, col_names], glm_coefs, y = training[, y_name])
-  x$mapping <- res
-  x$trained <- TRUE
-  x
+  step_lencode_glm_new(
+    terms = x$terms,
+    role = x$role,
+    trained = TRUE,
+    outcome = x$outcome,
+    mapping = res,
+    skip = x$skip,
+    id = x$id)
 }
 
 #' @importFrom stats as.formula glm binomial coef gaussian na.omit
@@ -166,11 +170,11 @@ map_glm_coef <- function(dat, mapping) {
 #' @importFrom recipes bake prep
 #' @importFrom purrr map
 #' @export
-bake.step_lencode_glm <- function(object, newdata, ...) {
+bake.step_lencode_glm <- function(object, new_data, ...) {
   for (col in names(object$mapping))
-    newdata[, col] <- map_glm_coef(newdata[, col], object$mapping[[col]])
+    new_data[, col] <- map_glm_coef(new_data[, col], object$mapping[[col]])
 
-  newdata
+  new_data
 }
 
 #' @importFrom recipes printer
@@ -187,6 +191,7 @@ print.step_lencode_glm <-
 #' @rdname step_lencode_glm
 #' @param x A `step_lencode_glm` object.
 #' @export
+#' @export tidy.step_lencode_glm
 tidy.step_lencode_glm <- function(x, ...) {
   if (is_trained(x)) {
     for(i in seq_along(x$mapping))
@@ -202,6 +207,7 @@ tidy.step_lencode_glm <- function(x, ...) {
       terms = term_names
     )
   }
+  res$id <- x$id
   res
 }
 

R/lme.R

@@ -28,6 +28,7 @@
 #'  the computations for subsequent operations
 #' @param trained A logical to indicate if the quantities for
 #'  preprocessing have been estimated.
+#' @param id A character string that is unique to this step to identify it.
 #' @return An updated version of `recipe` with the new step added
 #'  to the sequence of existing steps (if any). For the `tidy`
 #'  method, a tibble with columns `terms` (the selectors or
@@ -90,7 +91,8 @@ step_lencode_mixed <-
            outcome = NULL,
            options = list(verbose = 0),
            mapping = NULL,
-           skip = FALSE) {
+           skip = FALSE,
+           id = rand_id("lencode_bayes")) {
     if (is.null(outcome))
       stop("Please list a variable in `outcome`", call. = FALSE)
     add_step(
@@ -102,19 +104,14 @@ step_lencode_mixed <-
         outcome = outcome,
         options = options,
         mapping = mapping,
-        skip = skip
+        skip = skip,
+        id = id
       )
     )
   }
 
 step_lencode_mixed_new <-
-  function(terms = NULL,
-           role = NA,
-           trained = FALSE,
-           outcome = NULL,
-           options = NULL,
-           mapping = NULL,
-           skip = FALSE) {
+  function(terms, role, trained, outcome, options, mapping, skip, id) {
     step(
       subclass = "lencode_mixed",
       terms = terms,
@@ -123,7 +120,8 @@ step_lencode_mixed_new <-
       outcome = outcome,
       options = options,
       mapping = mapping,
-      skip = skip
+      skip = skip,
+      id = id
     )
   }
 
@@ -143,9 +141,16 @@ prep.step_lencode_mixed <- function(x, training, info = NULL, ...) {
   res <-
     map(training[, col_names], lme_coefs, y = training[[y_name]],
         x$options)
-  x$mapping <- res
-  x$trained <- TRUE
-  x
+  step_lencode_mixed_new(
+    terms = x$terms,
+    role = x$role,
+    trained = TRUE,
+    outcome = x$outcome,
+    options = x$options,
+    mapping = res,
+    skip = x$skip,
+    id = x$id
+  ) 
 }
 
 #' @importFrom stats as.formula binomial coef gaussian na.omit
@@ -204,11 +209,11 @@ map_lme_coef <- function(dat, mapping) {
 #' @importFrom recipes bake prep
 #' @importFrom purrr map
 #' @export
-bake.step_lencode_mixed <- function(object, newdata, ...) {
+bake.step_lencode_mixed <- function(object, new_data, ...) {
   for (col in names(object$mapping))
-    newdata[, col] <- map_lme_coef(newdata[, col], object$mapping[[col]])
+    new_data[, col] <- map_lme_coef(new_data[, col], object$mapping[[col]])
 
-  newdata
+  new_data
 }
 
 #' @importFrom recipes printer
@@ -225,6 +230,7 @@ print.step_lencode_mixed <-
 #' @rdname step_lencode_mixed
 #' @param x A `step_lencode_mixed` object.
 #' @export
+#' @export tidy.step_lencode_mixed
 tidy.step_lencode_mixed <- function(x, ...) {
   if (is_trained(x)) {
     for(i in seq_along(x$mapping))
@@ -240,6 +246,7 @@ tidy.step_lencode_mixed <- function(x, ...) {
       terms = term_names
     )
   }
+  res$id <- x$id
   res
 }