-
Notifications
You must be signed in to change notification settings - Fork 2k
/
frame.R
2441 lines (2217 loc) · 92.5 KB
/
frame.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#`
#` Frame and AST Nodes
#`
#` To conveniently and safely pass messages between R and H2O, this package
#` relies on S3 objects to capture and pass state. The end user will typically
#` never have to reason with these objects directly, as there are S3 accessor
#` methods provided for creating new objects.
#`
#` S3 Frame class objects are pointers to either data in an H2O cluster, or
#` potential data (future calculations) in the cluster. They are also classic
#` compiler AST Nodes (to hold future calculations). They are implemented with
#` simple R environment objects.
#`
#` Like AST Nodes in compilers all over, Frames build a simple DAG where the
#` nodes contain an operator and some outgoing edges. There is a GC finalizer
#` to delete the server-side copy of a Frame
#`
#`
#` === Frame/AST Node/environment Fields ===
#
#` E$op <- Operation or opcode that produces this Frame, a string
#
#` The combination of EVAL and ID fields determines the evaluation state:
#` EVAL is one of:
#` - TRUE : Node is evaluated, cluster has the ID, and an R GC finalizer will remove this temp ID
#` - FALSE: Node is evaluated, cluster has the ID, and the user has to explictly remove this permanent ID
#` - list of Nodes: Then further ID is one of:
#` - - missing: this Node is lazy and has never been evaluated
#` - - NA: this Node has been executed once, but no temp ID was made
#` - - String: this Node is-execution, with the given temp ID. Once execution has completed the EVAL field will be set to FALSE
#`
#` # A number of fields represent cached queries of an evaluated frame.
#` E$data <- A cached result; can be a scalar, or a R dataframe result holding
#` the first N (typically 10) rows and all cols of the frame
#` E$nrow <- the row count (total size, generally much larger than the local cached rows)
is.Frame <- function(fr) !missing(fr) && class(fr)[1]=="Frame"
chk.Frame <- function(fr) if( is.Frame(fr) ) fr else stop("must be a Frame")
#' Get back-end distributed key/value store id from a Frame.
#'
#' @param x A Frame
#' @return The id
#' @export
h2o.getId <- function(x) {
chk.Frame(x)
attr(x, "id")
}
.h2o.gc <- function() {
gc()
}
# Horrible internal shortcut to set our fields, using a more "normal"
# parameter order
.set <- function(x,name,value) attr(x,name) <- value
# GC Finalizer - called when GC collects a Frame Must be defined ahead of constructors.
.nodeFinalizer <- function(x) {
eval <- attr(x, "eval")
if( is.logical(eval) && eval ) {
#cat("=== Finalizer on ",attr(x, "id"),"\n")
.h2o.__remoteSend(paste0(.h2o.__DKV, "/", attr(x, "id")), method = "DELETE")
}
}
# Make a raw named data frame. The key will exist on the server, and will be
# the passed-in ID. Because it is named, it is not GCd. It is fully evaluated.
.newFrame <- function(op,id) {
stopifnot( is.character(id) )
node <- structure(new.env(parent = emptyenv()), class="Frame")
.set(node,"op",op)
.set(node,"id",id)
.set(node,"eval",FALSE) # User-managed lifetime
node
}
# A new lazy expression
.newExpr <- function(op,...) .newExprList(op,list(...))
.newExprList <- function(op,li) {
node <- structure(new.env(parent = emptyenv()), class="Frame")
.set(node,"op",op)
.set(node,"eval",li)
reg.finalizer(node, .nodeFinalizer, onexit=TRUE)
node
}
#' Data Frame Creation in H2O
#'
#' Creates a data frame in H2O with real-valued, categorical, integer, and binary columns specified by the user.
#'
#' @param key A string indicating the destination key. If empty, this will be auto-generated by H2O.
#' @param rows The number of rows of data to generate.
#' @param cols The number of columns of data to generate. Excludes the response column if \code{has_response = TRUE}.
#' @param randomize A logical value indicating whether data values should be randomly generated. This must be TRUE if either \code{categorical_fraction} or \code{integer_fraction} is non-zero.
#' @param value If \code{randomize = FALSE}, then all real-valued entries will be set to this value.
#' @param real_range The range of randomly generated real values.
#' @param categorical_fraction The fraction of total columns that are categorical.
#' @param factors The number of (unique) factor levels in each categorical column.
#' @param integer_fraction The fraction of total columns that are integer-valued.
#' @param integer_range The range of randomly generated integer values.
#' @param binary_fraction The fraction of total columns that are binary-valued.
#' @param binary_ones_fraction The fraction of values in a binary column that are set to 1.
#' @param missing_fraction The fraction of total entries in the data frame that are set to NA.
#' @param response_factors If \code{has_response = TRUE}, then this is the number of factor levels in the response column.
#' @param has_response A logical value indicating whether an additional response column should be pre-pended to the final H2O data frame. If set to TRUE, the total number of columns will be \code{cols+1}.
#' @param seed A seed used to generate random values when \code{randomize = TRUE}.
#' @return Returns a Frame object.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' hex <- h2o.createFrame(rows = 1000, cols = 100, categorical_fraction = 0.1,
#' factors = 5, integer_fraction = 0.5, integer_range = 1,
#' has_response = TRUE)
#' head(hex)
#' summary(hex)
#'
#' hex2 <- h2o.createFrame(rows = 100, cols = 10, randomize = FALSE, value = 5,
#' categorical_fraction = 0, integer_fraction = 0)
#' summary(hex2)
#' }
#' @export
h2o.createFrame <- function(key = "", rows = 10000, cols = 10, randomize = TRUE,
value = 0, real_range = 100, categorical_fraction = 0.2, factors = 100,
integer_fraction = 0.2, integer_range = 100, binary_fraction = 0.1,
binary_ones_fraction = 0.02, missing_fraction = 0.01, response_factors = 2,
has_response = FALSE, seed) {
if(!is.numeric(rows)) stop("`rows` must be a positive number")
if(!is.numeric(cols)) stop("`cols` must be a positive number")
if(!missing(seed) && !is.numeric(seed)) stop("`seed` must be a numeric value")
if(!is.logical(randomize)) stop("`randomize` must be TRUE or FALSE")
if(!is.numeric(value)) stop("`value` must be a numeric value")
if(!is.numeric(real_range)) stop("`real_range` must be a numeric value")
if(!is.numeric(categorical_fraction)) stop("`categorical_fraction` must be a numeric value")
if(!is.numeric(factors)) stop("`factors` must be a numeric value")
if(!is.numeric(integer_fraction)) stop("`integer_fraction` must be a numeric value")
if(!is.numeric(integer_range)) stop("`integer_range` must be a numeric value")
if(!is.numeric(binary_fraction)) stop("`binary_fraction` must be a numeric value")
if(!is.numeric(binary_ones_fraction)) stop("`binary_ones_fraction` must be a numeric value")
if(!is.numeric(missing_fraction)) stop("`missing_fraction` must be a numeric value")
if(!is.numeric(response_factors)) stop("`response_factors` must be a numeric value")
if(!is.logical(has_response)) stop("`has_response` must be a logical value")
.cframe.map <- c("key" = "dest")
parms <- lapply(as.list(match.call(expand.dots = FALSE)[-1L]), eval.parent, 2) # depth must be 2 in order to pop out of the lapply scope...
if(missing(key) || !is.character(key) || !nzchar(key))
parms$key = .key.make(prefix = "frame")
.key.validate(parms$key)
names(parms) <- lapply(names(parms), function(i) { if( i %in% names(.cframe.map) ) i <- .cframe.map[[i]]; i })
res <- .h2o.__remoteSend(.h2o.__CREATE_FRAME, method = "POST", .params = parms)
job_key <- res$key$name
dest_key <- res$dest$name
.h2o.__waitOnJob(job_key)
h2o.getFrame(dest_key)
}
#' Categorical Interaction Feature Creation in H2O
#'
#' Creates a data frame in H2O with n-th order interaction features between categorical columns, as specified by the user.
#'
#' @param data An Frame object containing the categorical columns.
#' @param destination_frame A string indicating the destination key. If empty, this will be auto-generated by H2O.
#' @param factors Factor columns (either indices or column names).
#' @param pairwise Whether to create pairwise interactions between factors (otherwise create one higher-order interaction). Only applicable if there are 3 or more factors.
#' @param max_factors Max. number of factor levels in pair-wise interaction terms (if enforced, one extra catch-all factor will be made)
#' @param min_occurrence Min. occurrence threshold for factor levels in pair-wise interaction terms
#' @return Returns a Frame object.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#'
#' # Create some random data
#' myframe = h2o.createFrame('framekey', rows = 20, cols = 5,
#' seed = -12301283, randomize = TRUE, value = 0,
#' categorical_fraction = 0.8, factors = 10, real_range = 1,
#' integer_fraction = 0.2, integer_range = 10,
#' binary_fraction = 0, binary_ones_fraction = 0.5,
#' missing_fraction = 0.2,
#' response_factors = 1)
#' # Turn integer column into a categorical
#' myframe[,5] <- as.factor(myframe[,5])
#' head(myframe, 20)
#'
#' # Create pairwise interactions
#' pairwise <- h2o.interaction(myframe, destination_frame = 'pairwise',
#' factors = list(c(1,2),c("C2","C3","C4")),
#' pairwise=TRUE, max_factors = 10, min_occurrence = 1)
#' head(pairwise, 20)
#' h2o.levels(pairwise,2)
#'
#' # Create 5-th order interaction
#' higherorder <- h2o.interaction(myframe, destination_frame = 'higherorder', factors = c(1,2,3,4,5),
#' pairwise=FALSE, max_factors = 10000, min_occurrence = 1)
#' head(higherorder, 20)
#'
#' # Limit the number of factors of the "categoricalized" integer column
#' # to at most 3 factors, and only if they occur at least twice
#' head(myframe[,5], 20)
#' trim_integer_levels <- h2o.interaction(myframe, destination_frame = 'trim_integers', factors = "C5",
#' pairwise = FALSE, max_factors = 3, min_occurrence = 2)
#' head(trim_integer_levels, 20)
#'
#' # Put all together
#' myframe <- h2o.cbind(myframe, pairwise, higherorder, trim_integer_levels)
#' myframe
#' head(myframe,20)
#' summary(myframe)
#' }
#' @export
h2o.interaction <- function(data, destination_frame, factors, pairwise, max_factors, min_occurrence) {
chk.Frame(data)
if(missing(factors)) stop("factors must be specified")
if(!is.logical(pairwise)) stop("pairwise must be a boolean value")
if(missing(max_factors)) stop("max_factors must be specified")
if(missing(min_occurrence)) stop("min_occurrence must be specified")
if (is.list(factors)) {
res <- lapply(factors, function(factor) h2o.interaction(data, destination_frame=NULL, factor, pairwise, max_factors, min_occurrence))
if (!missing(destination_frame)) {
old <- h2o.cbind(res)
new <- h2o.assign(old, destination_frame)
return(new)
} else {
return(h2o.cbind(res))
}
}
if(is.numeric(factors)) { factors <- colnames(data)[factors] }
if(is.numeric(factors)) stop("factors cannot be numeric value(s)")
if(is.null(factors)) stop("factors not found")
if(max_factors < 1) stop("max_factors cannot be < 1")
if(!is.numeric(max_factors)) stop("max_factors must be a numeric value")
if(min_occurrence < 1) stop("min_occurrence cannot be < 1")
if(!is.numeric(min_occurrence)) stop("min_occurrence must be a numeric value")
parms <- list()
if(missing(destination_frame) || !is.character(destination_frame) || !nzchar(destination_frame))
parms$dest = .key.make(prefix = "interaction")
.key.validate(parms$dest)
parms$source_frame <- attr(.eval.frame(data), "id")
parms$factor_columns <- .collapse.char(factors)
parms$pairwise <- pairwise
parms$max_factors <- max_factors
parms$min_occurrence <- min_occurrence
res <- .h2o.__remoteSend(page = 'Interaction', method = "POST", .params = parms)
job_key <- res$key$name
dest_key <- res$dest$name
.h2o.__waitOnJob(job_key)
h2o.getFrame(dest_key)
}
#' Replicate Elements of Vectors or Lists into H2O
#'
#' \code{h2o.rep} performs just as \code{rep} does. It replicates the values in
#' \code{x} in the H2O backend.
#'
#' @param x a vector (of any mode including a list) or a factor
#' @param length.out non negative integer. The desired length of the output
#' vector.
#' @return Creates a Frame vector of the same type as x
#' @export
h2o.rep_len <- function(x, length.out) {
if (length.out <= 0) NULL
else if( is.Frame(x) ) .newExpr("rep_len", chk.Frame(x), length.out)
else .newExpr("rep_len", x, length.out)
}
#' Inserting Missing Values to an H2O DataFrame
#'
#' *This is primarily used for testing*. Randomly replaces a user-specified fraction of
#' entries in a H2O dataset with missing values.
#'
#' @param data An Frame object representing the dataset.
#' @param fraction A number between 0 and 1 indicating the fraction of entries
#' to replace with missing.
#' @param seed A random number used to select which entries to replace with
#' missing values. Default of \code{seed = -1} will automatically
#' generate a seed in H2O.
#' @section WARNING: This will modify the original dataset. Unless this is intended,
#' this function should only be called on a subset of the original.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' irisPath <- system.file("extdata", "iris.csv", package = "h2o")
#' iris.hex <- h2o.importFile(path = irisPath)
#' summary(iris.hex)
#' irismiss.hex <- h2o.insertMissingValues(iris.hex, fraction = 0.25)
#' head(irismiss.hex)
#' summary(irismiss.hex)
#' }
#' @export
h2o.insertMissingValues <- function(data, fraction=0.1, seed=-1) {
parms = list()
parms$dataset <- attr(.eval.frame(data),"id") # Eager force evaluation
parms$fraction <- fraction
if( !missing(seed) )
parms$seed <- seed
json <- .h2o.__remoteSend(method = "POST", page = 'MissingInserter', .params = parms)
.h2o.__waitOnJob(json$key$name)
.flush.data(data) # Flush cache and return data
}
#' Split an H2O Data Set
#'
#' Split an existing H2O data set according to user-specified ratios.
#'
#' @param data An Frame object representing the dataste to split.
#' @param ratios A numeric value or array indicating the ratio of total rows
#' contained in each split. Must total up to less than 1.
#' @param destination_frames An array of frame IDs equal to the number of ratios
#' specified plus one.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' irisPath = system.file("extdata", "iris.csv", package = "h2o")
#' iris.hex = h2o.importFile(path = irisPath)
#' iris.split = h2o.splitFrame(iris.hex, ratios = c(0.2, 0.5))
#' head(iris.split[[1]])
#' summary(iris.split[[1]])
#' }
#' @export
h2o.splitFrame <- function(data, ratios = 0.75, destination_frames) {
params <- list()
params$dataset <- attr(.eval.frame(chk.Frame(data)), "id")
params$ratios <- .collapse(ratios)
if (!missing(destination_frames))
params$destination_frames <- .collapse.char(destination_frames)
res <- .h2o.__remoteSend(method="POST", "SplitFrame", .params = params)
job_key <- res$key$name
.h2o.__waitOnJob(job_key)
splits <- lapply(res$destination_frames, function(s) h2o.getFrame(s$name))
}
#'
#' Filter NA Columns
#'
#' @param data A dataset to filter on.
#' @param frac The threshold of NAs to allow per column (columns >= this threshold are filtered)
#' @export
h2o.filterNACols <- function(data, frac=0.2) {
(as.data.frame(.newExpr("filterNACols", data, frac)) + 1)[,1] # 0 to 1 based index
}
#' Cross Tabulation and Table Creation in H2O
#'
#' Uses the cross-classifying factors to build a table of counts at each combination of factor levels.
#'
#' @param x An Frame object with at most two columns.
#' @param y An Frame similar to x, or \code{NULL}.
#' @return Returns a tabulated Frame object.
#' @examples
#' library(h2o)
#' h2o.init()
#' prosPath <- system.file("extdata", "prostate.csv", package="h2o")
#' prostate.hex <- h2o.uploadFile(path = prosPath, destination_frame = "prostate.hex")
#' summary(prostate.hex)
#'
#' # Counts of the ages of all patients
#' head(h2o.table(prostate.hex[,3]))
#' h2o.table(prostate.hex[,3])
#'
#' # Two-way table of ages (rows) and race (cols) of all patients
#' head(h2o.table(prostate.hex[,c(3,4)]))
#' h2o.table(prostate.hex[,c(3,4)])
#' @export
h2o.table <- function(x, y = NULL) {
chk.Frame(x)
if( !is.null(y) ) chk.Frame(y)
if( is.null(y) ) .newExpr("table",x) else .newExpr("table",x,y)
}
#' @rdname h2o.table
#' @export
table.Frame <- h2o.table
#' H2O Median
#'
#' Compute the median of a Frame.
#'
#' @param x An Frame object.
#' @param na.rm a logical, indicating whether na's are omitted.
#' @examples
#' h2o.init()
#' prosPath <- system.file("extdata", "prostate.csv", package="h2o")
#' prostate.hex <- h2o.uploadFile(path = prosPath, destination_frame = "prostate.hex")
#' @export
h2o.median <- function(x, na.rm = TRUE) attr(.eval.frame(.newExpr("median",x,na.rm)), "data")
#' @rdname h2o.median
median.Frame <- h2o.median
#' Range of an H2O Column
#'
#' @param x An Frame object.
#' @param na.rm ignore missing values
#' @export
range <- function(x,na.rm = TRUE) {
if( !is.Frame(x) ) .Primitive("range")(x,na.rm)
else .newExpr("range",x,na.rm)
}
#' Cut H2O Numeric Data to Factor
#'
#' Divides the range of the H2O data into intervals and codes the values according to which interval they fall in. The
#' leftmost interval corresponds to the level one, the next is level two, etc.
#'
#' @param x An Frame object with numeric columns.
#' @param breaks A numeric vector of two or more unique cut points.
#' @param labels Labels for the levels of the resulting category. By default, labels are constructed sing "(a,b]"
#' interval notation.
#' @param include.lowest \code{Logical}, indicationg if an 'x[i]' equal to the lowest (or highest, for \code{right =
#' FALSE} 'breaks' value should be included
#' @param right /code{Logical}, indicating if the intervals should be closed on the right (opened on the left) or vice
#' versa.
#' @param dig.lab Integer which is used when labels are not given, determines the number of digits used in formatting
#' the break numbers.
#' @param ... Further arguments passed to or from other methods.
#' @return Returns an Frame object containing the factored data with intervals as levels.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' irisPath <- system.file("extdata", "iris_wheader.csv", package="h2o")
#' iris.hex <- h2o.uploadFile(path = irisPath, destination_frame = "iris.hex")
#' summary(iris.hex)
#'
#' # Cut sepal length column into intervals determined by min/max/quantiles
#' sepal_len.cut = cut(iris.hex$sepal_len, c(4.2, 4.8, 5.8, 6, 8))
#' head(sepal_len.cut)
#' summary(sepal_len.cut)
#' }
#' @export
h2o.cut <- function(x, breaks, labels = NULL, include.lowest = FALSE, right = TRUE, dig.lab = 3, ...) {
if (!is.numeric(breaks) || length(breaks) == 0L || !all(is.finite(breaks)))
stop("`breaks` must be a numeric vector")
.newExpr("cut", chk.Frame(x), breaks, labels, include.lowest, right, dig.lab)
}
#' @rdname h2o.cut
#' @export
cut.Frame <- h2o.cut
# `match` or %in% for Frame
#' Value Matching in H2O
#'
#' \code{match} and \code{\%in\%} return values similar to the base R generic
#' functions.
#'
#' @param x a categorical vector from an Frame object with
#' values to be matched.
#' @param table an R object to match \code{x} against.
#' @param nomatch the value to be returned in the case when no match is found.
#' @param incomparables a vector of calues that cannot be matched. Any value in
#' \code{x} matching a value in this vector is assigned the
#' \code{nomatch} value.
#' @seealso \code{\link[base]{match}} for base R implementation.
#' @examples
#' \donttest{
#' h2o.init()
#' hex <- as.h2o(iris)
#' h2o.match(hex[,5], c("setosa", "versicolor"))
#' }
#' @export
h2o.match <- function(x, table, nomatch = 0, incomparables = NULL) {
.newExpr("match", chk.Frame(x), table, nomatch, incomparables)
}
#' @rdname h2o.match
#' @export
match.Frame <- h2o.match
# %in% method
#' @rdname h2o.match
#' @export
`%in%` <- function(x,table) {
if( is.Frame(x) ) h2o.match(x,table,nomatch=0)
else base::`%in%`(x,table)
}
#' Remove Rows With NAs
#'
#' @rdname na.omit
#' @param object Frame object
#' @param ... Ignored
#' @export
na.omit.Frame <- function(object, ...) .newExpr("na.omit", object)
#' Compute DCT of an Frame
#'
#' Compute the Discrete Cosine Transform of every row in the Frame
#'
#' @param data An Frame object representing the dataset to transform
#' @param destination_frame A frame ID for the result
#' @param dimensions An array containing the 3 integer values for height, width, depth of each sample.
#' The product of HxWxD must total up to less than the number of columns.
#' For 1D, use c(L,1,1), for 2D, use C(N,M,1).
#' @param inverse Whether to perform the inverse transform
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' df <- h2o.createFrame(rows = 1000, cols = 8*16*24,
#' categorical_fraction = 0, integer_fraction = 0, missing_fraction = 0)
#' df1 <- h2o.dct(data=df, dimensions=c(8*16*24,1,1))
#' df2 <- h2o.dct(data=df1,dimensions=c(8*16*24,1,1),inverse=TRUE)
#' max(abs(df1-df2))
#'
#' df1 <- h2o.dct(data=df, dimensions=c(8*16,24,1))
#' df2 <- h2o.dct(data=df1,dimensions=c(8*16,24,1),inverse=TRUE)
#' max(abs(df1-df2))
#'
#' df1 <- h2o.dct(data=df, dimensions=c(8,16,24))
#' df2 <- h2o.dct(data=df1,dimensions=c(8,16,24),inverse=TRUE)
#' max(abs(df1-df2))
#' }
#' @export
h2o.dct <- function(data, destination_frame, dimensions, inverse=F) {
if(!is.logical(inverse)) stop("inverse must be a boolean value")
params <- list()
params$dataset <- attr(.eval.frame(chk.Frame(data)), "id")
params$dimensions <- .collapse(dimensions)
if (!missing(destination_frame))
params$destination_frame <- destination_frame
params$inverse <- inverse
res <- .h2o.__remoteSend(method="POST", h2oRestApiVersion = 99, "DCTTransformer", .params = params)
job_key <- res$key$name
.h2o.__waitOnJob(job_key)
h2o.getFrame(res$destination_frame$name)
}
#-----------------------------------------------------------------------------------------------------------------------
# Time & Date
#-----------------------------------------------------------------------------------------------------------------------
#' Convert Milliseconds to Years in H2O Datasets
#'
#' Convert the entries of a Frame object from milliseconds to years, indexed
#' starting from 1900.
#'
# is this still true?
#' This method calls the function of the MutableDateTime class in Java.
#' @param x An Frame object.
#' @return A Frame object containig the entries of \code{x} converted to years
#' starting from 1900, e.g. 69 corresponds to the year 1969.
#' @seealso \code{\link{h2o.month}}
#' @export
h2o.year <- function(x) .newExpr("-",.newExpr("year", chk.Frame(x)),1900)
#' Convert Milliseconds to Months in H2O Datasets
#'
#' Converts the entries of a Frame object from milliseconds to months (on a 1 to
#' 12 scale).
#'
#' @param x An Frame object.
#' @return A Frame object containing the entries of \code{x} converted to months of
#' the year.
#' @seealso \code{\link{h2o.year}}
#' @export
h2o.month <- function(x) .newExpr("month", chk.Frame(x))
#' Convert Milliseconds to Week of Week Year in H2O Datasets
#'
#' Converts the entries of a Frame object from milliseconds to weeks of the week
#' year (starting from 1).
#'
#' @param x An Frame object.
#' @return A Frame object containing the entries of \code{x} converted to weeks of
#' the week year.
#' @seealso \code{\link{h2o.month}}
#' @export
h2o.week <- function(x) .newExpr("week", chk.Frame(x))
#' Convert Milliseconds to Day of Month in H2O Datasets
#'
#' Converts the entries of a Frame object from milliseconds to days of the month
#' (on a 1 to 31 scale).
#'
#' @param x An Frame object.
#' @return A Frame object containing the entries of \code{x} converted to days of
#' the month.
#' @seealso \code{\link{h2o.month}}
#' @export
h2o.day <- function(x) .newExpr("day", chk.Frame(x))
#' Convert Milliseconds to Day of Week in H2O Datasets
#'
#' Converts the entries of a Frame object from milliseconds to days of the week
#' (on a 0 to 6 scale).
#'
#' @param x An Frame object.
#' @return A Frame object containing the entries of \code{x} converted to days of
#' the week.
#' @seealso \code{\link{h2o.day}, \link{h2o.month}}
#' @export
h2o.dayOfWeek <- function(x) .newExpr("dayOfWeek", chk.Frame(x))
#' Convert Milliseconds to Hour of Day in H2O Datasets
#'
#' Converts the entries of a Frame object from milliseconds to hours of the day
#' (on a 0 to 23 scale).
#'
#' @param x An Frame object.
#' @return A Frame object containing the entries of \code{x} converted to hours of
#' the day.
#' @seealso \code{\link{h2o.day}}
#' @export
h2o.hour <- function(x) .newExpr("hour", chk.Frame(x))
#' @rdname h2o.year
#' @export
year <- function(x) UseMethod('year', x)
#' @rdname h2o.year
#' @export
year.Frame <- h2o.year
#' @rdname h2o.month
#' @export
month <- function(x) UseMethod('month', x)
#' @rdname h2o.month
#' @export
month.Frame <- h2o.month
#' @rdname h2o.week
#' @export
week <- function(x) UseMethod('week', x)
#' @rdname h2o.week
#' @export
week.Frame <- h2o.week
#' @rdname h2o.day
#' @export
day <- function(x) UseMethod('day', x)
#' @rdname h2o.day
#' @export
day.Frame <- h2o.day
#' @rdname h2o.dayOfWeek
#' @export
dayOfWeek <- function(x) UseMethod('dayOfWeek', x)
#' @rdname h2o.dayOfWeek
#' @export
dayOfWeek.Frame <- h2o.dayOfWeek
#' @rdname h2o.hour
#' @export
hour <- function(x) UseMethod('hour', x)
#' @rdname h2o.hour
#' @export
hour.Frame <- h2o.hour
#' @export
as.Date.Frame <- function(x, format, ...) {
if(!is.character(format)) stop("format must be a string")
.newExpr("as.Date", chk.Frame(x), .quote(format), ...)
}
#' Set the Time Zone on the H2O Cloud
#'
#' @param tz The desired timezone.
#' @export
h2o.setTimezone <- function(tz) .fetch.data(.newExpr("setTimeZone",.quote(tz)),1000)
#' Get the Time Zone on the H2O Cloud
#' Returns a string
#'
#' @export
h2o.getTimezone <- function() .fetch.data(.newExpr("getTimeZone"),1000)
#' List all of the Time Zones Acceptable by the H2O Cloud.
#'
#' @export
h2o.listTimezones <- function() .fetch.data(.newExpr("listTimeZones"),1000)
#' Produce a Vector of Random Uniform Numbers
#'
#' Creates a vector of random uniform numbers equal in length to the length of the specified H2O
#' dataset.
#'
#' @param x An Frame object.
#' @param seed A random seed used to generate draws from the uniform distribution.
#' @return A vector of random, uniformly distributed numbers. The elements are between 0 and 1.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' prosPath = system.file("extdata", "prostate.csv", package="h2o")
#' prostate.hex = h2o.importFile(path = prosPath, destination_frame = "prostate.hex")
#' s = h2o.runif(prostate.hex)
#' summary(s)
#'
#' prostate.train = prostate.hex[s <= 0.8,]
#' prostate.train = h2o.assign(prostate.train, "prostate.train")
#' prostate.test = prostate.hex[s > 0.8,]
#' prostate.test = h2o.assign(prostate.test, "prostate.test")
#' nrow(prostate.train) + nrow(prostate.test)
#' }
#' @export
h2o.runif <- function(x, seed = -1) {
if (!is.numeric(seed) || length(seed) != 1L || !is.finite(seed)) stop("`seed` must be an integer >= 0")
if (seed == -1) seed <- floor(runif(1,1,.Machine$integer.max*100))
.newExpr("h2o.runif", chk.Frame(x), seed)
}
#' Check Frame columns for factors
#'
#' Determines if any column of an Frame object contains categorical data.
#'
#' @name h2o.anyFactor
#' @param x An \code{Frame} object.
#' @return Returns a logical value indicating whether any of the columns in \code{x} are factors.
#' @examples
#' \donttest{
#' library(h2o)
#' h2o.init()
#' irisPath <- system.file("extdata", "iris_wheader.csv", package="h2o")
#' iris.hex <- h2o.importFile(path = irisPath)
#' h2o.anyFactor(iris.hex)
#' }
#' @export
h2o.anyFactor <- function(x) .newExpr("any.factor", chk.Frame(x))
#-----------------------------------------------------------------------------------------------------------------------
# Overloaded Base R Methods
#-----------------------------------------------------------------------------------------------------------------------
#-----------------------------------------------------------------------------------------------------------------------
# Slicing
#-----------------------------------------------------------------------------------------------------------------------
# Convert to Currents number-list syntax
.num.list <- function(nl) paste0('[',paste0(nl,collapse=" "),']')
# Convert to Currents string-list syntax
.quote <- function(x) paste0('"',x,'"')
.str.list <- function(sl) paste0('[',paste0('"',sl,'"',collapse=" "),']')
# Convert a row or column selector to zero-based numbering and return a string
.row.col.selector <- function( sel, raw_sel=NULL, envir=NULL ) {
if( !is.symbol(sel) && is.language(sel) && sel[[1]] == ":" ) {
sub <- if( eval(sel[[2]], envir=envir) < 0 ) 0 else 1L
s <- paste0( "[", eval(sel[[2]], envir=envir) - sub, ":", abs(eval(sel[[3]], envir=envir) - eval(sel[[2]], envir=envir))+1L, "]")
return( s )
}
sel <- if( !is.null(raw_sel) ) raw_sel else eval(sel)
if( is.numeric(sel) ) { # number list for column selection; zero based
sel2 <- lapply(sel,function(x) if( x==0 ) stop("Cannot select row or column 0") else if( x > 0 ) x-1 else x)
.num.list(sel2)
} else {
if( is.null(sel) ) "[]" # Empty selector
else as.character(sel)
}
}
#' Extract or Replace Parts of an Frame Object
#'
#' Operators to extract or replace parts of Frame objects.
#'
#' @name Frame-Extract
NULL
#' @aliases [,Frame-method
#' @rdname Frame-Extract
#' @param data object from which to extract element(s) or in which to replace element(s).
#' @param row index specifying row element(s) to extract or replace. Indices are numeric or
#' character vectors or empty (missing) or will be matched to the names.
#' @param col index specifying column element(s) to extract or replace.
#' @param drop Unused
#' @export
`[.Frame` <- function(data,row,col,drop=TRUE) {
chk.Frame(data)
# This function is called with a huge variety of argument styles
# Here's the breakdown:
# Style Type #args Description
# df[] - na na 2 both missing, identity with df
# df["colname"] - c na 2 single column by name, df$colname
# df[3] - X na 2 if ncol > 1 then column else row
# df[,] - na na 3 both missing, identity with df
# df[2,] - r na 3 constant row, all cols
# df[1:150,] - r na 3 selection of rows, all cols
# df[,3] - na c 3 constant column
# df[,1:10] - na c 3 selection of columns
# df[,"colname"] - na c 3 single column by name
# df[2,"colname"]- r c 3 row slice and column-by-name
# df[2,3] - r c 3 single element
# df[1:150,1:10] - r c 3 rectangular slice
# df[a<b,] - f na 3 boolean row slice
# df[a<b,c] - f c 3 boolean row slice
is1by1 <- !missing(col) && !missing(row) && !is.Frame(row) && length(col) == 1 && length(row) == 1
if( nargs() == 2 && # Only row, no column; nargs==2 distinguishes "df[2,]" (row==2) from "df[2]" (col==2)
# is.char tells cars["cylinders"], or if there are multiple columns.
# Single column with numeric selector is row: car$cylinders[100]
(is.character(row) || ncol(data) > 1) ) {
# Row is really column: cars[3] or cars["cylinders"] or cars$cylinders
col <- row
row <- NA
}
if( !missing(col) ) { # Have a column selector?
if( is.logical(col) ) { # Columns by boolean choice
col <- which(col) # Pick out all the TRUE columns by index
} else if( is.character(col) ) { # Columns by name
assign("idx", match(col,colnames(data))) # Match on name
if( any(is.na(idx)) ) stop(paste0("No column '",col,"' found in ",paste(colnames(data),collapse=",")))
col <- idx
}
idx <- .row.col.selector(col,envir=parent.frame()) # Generic R expression
assign("data",.newExpr("cols",data,idx)) # Column selector
}
# Have a row selector?
if( !missing(row) && (is.Frame(row) || !is.na(row)) ) {
if( !is.Frame(row) ) # Generic R expression
row <- .row.col.selector(substitute(row), row,envir=parent.frame())
assign("data",.newExpr("rows",data,row)) # Row selector
}
if( is1by1 ) .fetch.data(data,1)[[1]]
else data
}
#' @rdname Frame-Extract
#' @param x An H2O Frame
#' @param name a literal character string or a name (possibly backtick quoted).
#' @export
`$.Frame` <- function(x, name) { x[[name, exact = FALSE]] }
#' @rdname Frame-Extract
#' @param i index
#' @param exact controls possible partial matching of \code{[[} when extracting
#' a character
#' @export
`[[.Frame` <- function(x, i, exact = TRUE) {
if( missing(i) ) return(x)
if( length(i) > 1L ) stop("`[[` can only select one column")
if( is.character(i)) {
if( exact ) i <- match(i, colnames(x))
else i <- pmatch(i, colnames(x))
}
if( is.na(i) ) NULL
else x[,i]
}
#
# Overload Assignment!
#
# Trying to remove excessive temp generation, by having the R interpreter tell
# H2O that some computation may be used, or not. If the expression is only
# ever used once, then no temp is needed and the cluster can optimize the
# lifetime. If the temp *may* be used again, the cluster needs a temp for
# the reuse, or else the computation needs to be "pure" and re-executed.
#
# After many many attempts, I think it's not reasonably possible to track
# lifetimes in R via assignment overload. There are too many other paths
# that extend lifetimes that all must be caught (including, but not limited
# to: c, list, <-, =, and the *apply series)
# Internal recursive printer
.pfr <- function(x) {
if( is.list(res<- attr(x,"eval")) )
res <- paste0("(",attr(x, "op")," ",paste(sapply( attr(x,"eval"), function(child) { if( is.Frame(child) ) .pfr(child) else child }),collapse=" "),")")
paste0( attr(x, "id"), ":=", res)
}
# Pretty print the reachable execution DAG from this Frame, withOUT evaluating it
pfr <- function(x) { chk.Frame(x); .pfr(x) }
.eval.impl <- function(x) {
dat <- attr(x, "data")
id <- attr(x, "id")
if( !is.null(dat) ) return( if( is.data.frame(dat) ) id else dat ) # Data already computed and cached
if( !is.null( id) && !is.na(id) ) return( id ) # Data already computed under ID
# Build the eval expression
stopifnot(is.list(attr(x, "eval")))
res <- paste(sapply( attr(x, "eval"), function(child) {
if( is.Frame(child) ) .eval.impl(child) # recurse
else if( is.numeric( child) && length(child) > 1L ) .num.list(child) # [ numberz ] TODO: sup with those NaNs tho
else if( is.character(child) && length(child) > 1L ) .str.list(child) # [ stringz ]
else child # base
}),collapse=" ")
res <- paste0("(",attr(x, "op")," ",res,")")
# First exec: ID is missing, convert to NA
# 2nd exec: ID is NA, convert to unique string
# 3rd exec: there is no 3rd exec, just use the ID string
if( is.null(id) ) .set(x,"id",NA) # 1st exec: missing->NA
else { # 2nd exec: NA-> tmp name
.set(x,"id", id <- .key.make("RTMP")) # Flag as code-emitted by assigning the cluster name
res <- paste0("(tmp= ",id," ",res,")")
}
res
}
.clear.impl <- function(x) {
if( !is.Frame(x) ) return()
eval <- attr(x, "eval")
if( !is.list(eval) ) { stopifnot(is.character( attr(x, "id") )); return() }
lapply(eval, function(child) .clear.impl(child))
if( is.character( attr(x, "id")) )
.set(x,"eval",TRUE) # GC-able temp
}
# Evaluate this Frame on demand.
#
# Because of GC, this algo requires 2 passes over the DAG. The first pass
# builds the expression string - but it cannot let any of the sub-parts go
# dead, lest GC delete frames on last use... before the expression string is
# shipped over the wire. During the 2nd pass the internal DAG pointers are
# wiped out, and allowed to go dead (hence can be nuked by GC).
.eval.frame <- function(x) {
chk.Frame(x)
if( !is.character( attr(x, "id")) ) {
exec_str <- .eval.impl(x)
if( !is.character( attr(x, "id")) ) # Top-level gets a name always
.set(x,"id", id <- .key.make("RTMP"))
# Execute the AST on H2O
#print(paste0("EXPR: ",exec_str))
res <- .h2o.__remoteSend(.h2o.__RAPIDS, h2oRestApiVersion = 99, ast=exec_str, id= attr(x, "id"), method = "POST")
if( !is.null(res$error) ) stop(paste0("Error From H2O: ", res$error), call.=FALSE)
if( !is.null(res$scalar) ) {
y <- res$scalar
if( y=="TRUE" ) y <- TRUE
else if( y=="FALSE" ) y <- FALSE
.set(x,"data", y)
}
# Now clear all internal DAG nodes, allowing GC to reclaim them
.clear.impl(x)
.fetch.data(x,1) #trigger a cache update if needed
# Enable this GC to trigger rapid R GC cycles, and rapid R clearing of
# temps... to help debug GC issues.
#.h2o.gc()
}
x
}
#' S3 Group Generic Functions for H2O
#'
#' Methods for group generic functions and H2O objects.
#'
#' @rdname Frame
#' @param e1 object
#' @param e2 object
#' @export
Ops.Frame <- function(e1,e2)
.newExpr(.Generic,
if( is.character(e1) ) .quote(e1) else e1,
if( is.character(e2) ) .quote(e2) else e2)
#' @rdname Frame
#' @param x object
#' @export
Math.Frame <- function(x) .newExpr(.Generic,x)
#' @rdname Frame
#' @param y object
#' @export
Math.Frame <- function(x,y) .newExpr(.Generic,x,y)
#' @rdname Frame
#' @param ... Further arguments passed to or from other methods.
#' @export
Math.Frame <- function(x,...) .newExprList(.Generic,list(x,...))
#' @rdname Frame
#' @param na.rm logical. whether or not missing values should be removed
#' @export
Summary.Frame <- function(x,...,na.rm) {
if( na.rm ) stop("na.rm versions not impl")
# Eagerly evaluation, to produce a scalar
res <- attr(.eval.frame(.newExprList(.Generic,list(x,...))), "data")
if( .Generic=="all" ) as.logical(res) else res
}
#' @rdname Frame
#' @export
`!.Frame` <- function(x) .newExpr("!!",x)
#' @rdname Frame
#' @export
is.na.Frame <- function(x) .newExpr("is.na", x)
#' @rdname Frame
#' @export
t.Frame <- function(x) .newExpr("t",x)
#' @rdname Frame
#' @export
log <- function(x, ...) {
if( !is.Frame(x) ) .Primitive("log")(x)