Merge branch 'master' into integer_overflow

Author: mb706

DESCRIPTION

@@ -62,6 +62,7 @@ Depends:
 Imports:
     backports,
     checkmate,
+    cli,
     data.table,
     digest,
     lgr,

NAMESPACE

@@ -229,6 +229,7 @@ if (getRversion() >= "4.3.0") S3method(chooseOpsMethod,CnfAtom)
 if (getRversion() >= "4.3.0") S3method(chooseOpsMethod,CnfClause)
 if (getRversion() >= "4.3.0") S3method(chooseOpsMethod,CnfFormula)
 import(checkmate)
+import(cli)
 import(data.table)
 import(mlr3)
 import(mlr3misc)

NEWS.md

@@ -1,12 +1,13 @@
 # mlr3pipelines 0.9.0-9000
 
-* Compatibility with new testthat version 3.3.0
+* Pretty-printing some info using the `cli` package now.
 * Fix: Added internal workaround for `PipeOpNMF` attaching `Biobase`, `BiocGenerics`, and `generics` to the search path during training, prediction or when printing its `$state`.
 * feat: allow dates in datefeatures pipe op and use data.table for date feature generation.
 * Added support for internal validation tasks to `PipeOpFeatureUnion`.
 * feat: `PipeOpLearnerCV` can reuse the cross-validation models during prediction by averaging their outputs (`resampling.predict_method = "cv_ensemble"`).
 * feat: `PipeOpRegrAvg` gets new `se_aggr` and `se_aggr_rho` hyperparameters and now allows various forms of SE aggregation.
 * Fix: `PipeOpRemoveConstants` now avoids integer overflow when evaluating relative tolerances for near-`integer.max` data.
+* Compatibility with new testthat version 3.3.0
 
 # mlr3pipelines 0.9.0
 

R/Graph.R

@@ -414,15 +414,25 @@ Graph = R6Class("Graph",
           scc = self$edges[, list(sccssors = paste(unique(dst_id), collapse = ",")), by = list(ID = src_id)]
           lines = scc[prd[lines, on = "ID"], on = "ID"][, c("ID", "State", "sccssors", "prdcssors")]
           lines[is.na(lines)] = ""
-          catf("Graph with %s PipeOps:", nrow(lines))
+          cat_cli(cli_h1("Graph with {nrow(lines)} PipeOps:"))
           ## limit column width ##
 
           outwidth = getOption("width") %??% 80  # output width we want (default 80)
           colwidths = map_int(lines, function(x) max(nchar(x), na.rm = TRUE))  # original width of columns
           collimit = calculate_collimit(colwidths, outwidth)
+
           opts = options(datatable.prettyprint.char = collimit)
           on.exit(options(opts), add = TRUE)
           print(lines, row.names = FALSE)
+
+          is_sequential = all(table(self$edges$src_id) <= 1) && all(table(self$edges$dst_id) <= 1)
+          if(is_sequential) {
+            ppunit = paste0(self$ids(), collapse = " -> ")
+            pp = paste0(c("<INPUT>", ppunit, "<OUTPUT>"), collapse = " -> ")
+          } else {
+            pp = "non-sequential"
+          }
+          cat_cli(cli_h3("Pipeline: {.strong {pp}}"))
         } else {
           cat("Empty Graph.\n")
         }

R/GraphLearner.R

@@ -312,6 +312,18 @@ GraphLearner = R6Class("GraphLearner", inherit = Learner,
     },
     plot = function(html = FALSE, horizontal = FALSE, ...) {
       private$.graph$plot(html = html, horizontal = horizontal, ...)
+    },
+    print = function() {
+      super$print(self)
+
+      is_sequential = all(table(self$edges$src_id) <= 1) && all(table(self$edges$dst_id) <= 1)
+      if(is_sequential) {
+        ppunit = paste0(self$ids(), collapse = " -> ")
+        pp = paste0(c("<INPUT>", ppunit, "<OUTPUT>"), collapse = " -> ")
+      } else {
+        pp = "non-sequential"
+      }
+      cat_cli(cli_h3("Pipeline: {.strong {pp}}"))
     }
   ),
   active = list(

R/PipeOp.R

@@ -278,18 +278,23 @@ PipeOp = R6Class("PipeOp",
 
     print = function(...) {
       type_table_printout = function(table) {
-        strings = do.call(sprintf, cbind(fmt = "%s`[%s,%s]", table[, c("name", "train", "predict")]))
-        strings = strwrap(paste(strings, collapse = ", "), indent = 2, exdent = 2)
-        if (length(strings) > 6) {
-          strings = c(strings[1:5], sprintf("  [... (%s lines omitted)]", length(strings) - 5))
+        print(head(table, 5L), row.names = FALSE, print.keys = FALSE)
+        if (nrow(table) > 5L) {
+          catf("[...] (%i rows omitted)", nrow(table) - 5L)
         }
-        gsub("`", " ", paste(strings, collapse = "\n"))
       }
 
-      catf("PipeOp: <%s> (%strained)", self$id, if (self$is_trained) "" else "not ")
-      catf("values: <%s>", as_short_string(self$param_set$values))
-      catf("Input channels <name [train type, predict type]>:\n%s", type_table_printout(self$input))
-      catf("Output channels <name [train type, predict type]>:\n%s", type_table_printout(self$output))
+      msg_h = if (self$is_trained) "" else "not "
+      cat_cli({
+        cli_h1("PipeOp {.cls {self$id}}: {msg_h}trained")
+        cli_text("Values: {as_short_string(self$param_set$values)}")
+        cli_h3("{.strong Input channels:}")
+      })
+      type_table_printout(self$input)
+      cat_cli({
+        cli_h3("{.strong Output channels:}")
+      })
+      type_table_printout(self$output)
     },
 
     train = function(input) {

R/PipeOpFixFactors.R

@@ -45,12 +45,30 @@
 #' @section Methods:
 #' Only methods inherited from [`PipeOpTaskPreprocSimple`]/[`PipeOpTaskPreproc`]/[`PipeOp`].
 #'
+#' @examples
+#' library("mlr3")
+#'
+#' # Reduced task with no entries for the installment_rate < 20 is defined
+#' task = tsk("german_credit")
+#' rows = task$row_ids[task$data()[, installment_rate != "< 20"]]
+#' reduced_task = task$clone(deep = TRUE)$filter(rows)
+#' levels(reduced_task$data()$installment_rate)
+#'
+#' # PipeOp is trained on the reduced task
+#' po = po("fixfactors")
+#' processed_task = preproc(reduced_task, po)
+#' levels(processed_task$data()$installment_rate)
+#' summary(processed_task$data()$installment_rate)
+#'
+#' predicted_task = preproc(task, po, predict = TRUE)
+#'
+#' # Predictions are made on the task without any missing data
+#' levels(predicted_task$data()$installment_rate)
+#' summary(predicted_task$data()$installment_rate)
 #' @family PipeOps
 #' @template seealso_pipeopslist
 #' @include PipeOpTaskPreproc.R
 #' @export
-#' @examples
-#' library("mlr3")
 PipeOpFixFactors = R6Class("PipeOpFixFactors",
   inherit = PipeOpTaskPreprocSimple,
   public = list(

R/PipeOpTrafo.R

@@ -221,6 +221,37 @@ PipeOpTargetTrafo = R6Class("PipeOpTargetTrafo",
 #' @section Methods:
 #' Only methods inherited from [`PipeOp`].
 #'
+#' @examplesIf requireNamespace("rpart")
+#' library(mlr3)
+#' task = tsk("boston_housing")
+#' po = PipeOpTargetMutate$new("logtrafo", param_vals = list(
+#'   trafo = function(x) log(x, base = 2),
+#'   inverter = function(x) list(response = 2 ^ x$response))
+#' )
+#' # Note that this example is ill-equipped to work with
+#' # `predict_type == "se"` predictions.
+#'
+#' po$train(list(task))
+#' po$predict(list(task))
+#'
+#' g = Graph$new()
+#' g$add_pipeop(po)
+#' g$add_pipeop(LearnerRegrRpart$new())
+#' g$add_pipeop(PipeOpTargetInvert$new())
+#' g$add_edge(src_id = "logtrafo", dst_id = "targetinvert",
+#'   src_channel = 1, dst_channel = 1)
+#' g$add_edge(src_id = "logtrafo", dst_id = "regr.rpart",
+#'   src_channel = 2, dst_channel = 1)
+#' g$add_edge(src_id = "regr.rpart", dst_id = "targetinvert",
+#'   src_channel = 1, dst_channel = 2)
+#'
+#' g$train(task)
+#' g$predict(task)
+#'
+#' #syntactic sugar using ppl():
+#' tt = ppl("targettrafo", graph = PipeOpLearner$new(LearnerRegrRpart$new()))
+#' tt$param_set$values$targetmutate.trafo = function(x) log(x, base = 2)
+#' tt$param_set$values$targetmutate.inverter = function(x) list(response = 2 ^ x$response)
 #' @family PipeOps
 #' @template seealso_pipeopslist
 #' @include PipeOp.R

R/multiplicity.R

@@ -56,7 +56,7 @@ print.Multiplicity = function(x, ...) {
   if (!length(x)) {
     cat("Empty Multiplicity.\n")
   } else {
-    cat("Multiplicity:\n")
+    cli_h2("Multiplicity:")
     print(unclass(x), ...)
   }
 }

R/zzz.R

@@ -1,5 +1,6 @@
 #' @import data.table
 #' @import checkmate
+#' @import cli
 #' @import mlr3
 #' @import paradox
 #' @import mlr3misc