-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
1208 lines (1061 loc) · 32.7 KB
/
index.html
File metadata and controls
1208 lines (1061 loc) · 32.7 KB
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
<!DOCTYPE html>
<html>
<head>
<title>Large Data and R</title>
<meta charset="utf-8">
<meta name="description" content="Large Data and R">
<meta name="author" content="Miguel Vazquez-Prada, September 2013.">
<meta name="generator" content="slidify" />
<meta name="apple-mobile-web-app-capable" content="yes">
<meta http-equiv="X-UA-Compatible" content="chrome=1">
<link rel="stylesheet" href="libraries/frameworks/io2012/css/default.css" media="all" >
<link rel="stylesheet" href="libraries/frameworks/io2012/css/phone.css"
media="only screen and (max-device-width: 480px)" >
<link rel="stylesheet" href="libraries/frameworks/io2012/css/slidify.css" >
<link rel="stylesheet" href="libraries/highlighters/highlight.js/css/tomorrow_mig.css" />
<base target="_blank"> <!-- This amazingness opens all links in a new tab. --> <link rel=stylesheet href="libraries/widgets/bootstrap/css/bootstrap.css"></link>
<link rel=stylesheet href="libraries/widgets/interactive/css/aceeditor.css"></link>
<link rel=stylesheet href="libraries/widgets/nvd3/css/nv.d3.css"></link>
<link rel=stylesheet href="libraries/widgets/nvd3/css/rNVD3.css"></link>
<link rel=stylesheet href="./assets/css/ribbons.css"></link>
<!-- Grab CDN jQuery, fall back to local if offline -->
<script src="http://ajax.aspnetcdn.com/ajax/jQuery/jquery-1.7.min.js"></script>
<script>window.jQuery || document.write('<script src="libraries/widgets/quiz/js/jquery.js"><\/script>')</script>
<script data-main="libraries/frameworks/io2012/js/slides"
src="libraries/frameworks/io2012/js/require-1.0.8.min.js">
</script>
<script src="libraries/widgets/nvd3/js/jquery-1.8.2.min.js"></script>
<script src="libraries/widgets/nvd3/js/d3.v3.min.js"></script>
<script src="libraries/widgets/nvd3/js/nv.d3.min-new.js"></script>
<script src="libraries/widgets/nvd3/js/fisheye.js"></script>
</head>
<body style="opacity: 0">
<slides class="layout-widescreen">
<!-- LOGO SLIDE -->
<slide class="title-slide segue nobackground">
<hgroup class="auto-fadein">
<h1>Large Data and R</h1>
<h2>how to deal with large data in R with no memory errors and efficient code</h2>
<p>Miguel Vazquez-Prada, September 2013.<br/>Center for Epidemiology and Screening, University of Copenhagen.</p>
</hgroup>
<article></article>
</slide>
<!-- SLIDES -->
<slide class="" id="slide-1" style="background:#EDE0CF;">
<hgroup>
<h2>R</h2>
</hgroup>
<article data-timings="">
<p><strong>R</strong> is a system for statistical computation and graphics.
It consists of a language plus a run-time environment with graphics, a debugger, access to certain system functions, and the ability to run programs stored in script files.</p>
<p><strong>R</strong> was initially written by Ross Ihaka and Robert Gentleman at the Department of Statistics of the University of Auckland in Auckland, New Zealand. A large group of individuals has contributed to R by sending code and bug reports.
Since mid-1997 there has been a core group (the "R Core Team") who can modify the R source code archive.</p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-2" style="background:;">
<hgroup>
<h2>Why R ??</h2>
</hgroup>
<article data-timings="">
<blockquote>
<p><strong>R</strong> is dynamic, to say the least.
More precisely, it is organic, with new functionality and add-on packages appearing constantly.
And because of its open-source nature and free availability,
R is quickly becoming the software of choice for statistical analysis in a variety of fields. </p>
</blockquote>
<p><a href="http://cran.r-project.org/web/packages/HSAUR/vignettes/Ch_introduction_to_R.pdf" title="R intro">A Handbook of Statistical Analysis Using R</a></p>
<p><img src="assets/img/R.png" alt="R"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-3" style="background:;">
<hgroup>
<h2>Why R?</h2>
</hgroup>
<article data-timings="">
<ol class = "build incremental">
<li>OPEN SOURCE </li>
<li>it is used in an increasing number of research centers and media. </li>
<li>with some patience, you can do almost ANYTHING with your data. </li>
<li>is similar to other high performance languages, like C++ , python, etc. </li>
<li>it is cross platform and can be extremely efficient. </li>
<li>integration with other languages (C++, etc.).</li>
<li>large amount of specific packages-</li>
<li>great graphics capabilities.</li>
<li>very dynamic users community</li>
<li><strong>it is the FUTURE...</strong> </li>
</ol>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-4" style="background:;">
<hgroup>
<h2>SAS VS R</h2>
</hgroup>
<article data-timings="">
<table><thead>
<tr>
<th>SAS</th>
<th>R</th>
</tr>
</thead><tbody>
<tr>
<td>Payed</td>
<td>free</td>
</tr>
<tr>
<td>Medium performance</td>
<td>Variable performance</td>
</tr>
<tr>
<td>No memory issues</td>
<td>Possible memory issues</td>
</tr>
<tr>
<td>Stable workflow</td>
<td>Custom workflow</td>
</tr>
<tr>
<td>Difficult to expand</td>
<td>Expandable, full amount of packages</td>
</tr>
<tr>
<td>Poor graphics</td>
<td>Great graphics</td>
</tr>
<tr>
<td>Fixed documentation</td>
<td>Disperse, but very active community</td>
</tr>
<tr>
<td><em>Recommended to:</em></td>
<td><em>Recommended to:</em></td>
</tr>
<tr>
<td>Linear, non-programmers</td>
<td>non-linear, custom and graphics</td>
</tr>
</tbody></table>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-5" style="background:;">
<hgroup>
<h2>MOTIVATION</h2>
</hgroup>
<article data-timings="">
<p><strong>The "approach":</strong> </p>
<ul>
<li>Science as an open and accessible social development. </li>
<li>World-wide sharing and support</li>
<li>Freedom to customize </li>
</ul>
<p><img src="assets/img/globals.jpg" alt="R"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-6" style="background:;">
<hgroup>
<h2>MOTIVATION 2</h2>
</hgroup>
<article data-timings="">
<p><strong>R</strong> is an open source programming language that is widely used in statistical computations.
More and more researchers and Universities are using R as the main statistical analysis language. </p>
<p><strong>R</strong> is one of the best options to use in statistical analysis.</p>
<p>BUT, </p>
<p><strong>R</strong> faces many problems while using Large Data Sets.
And that can make R a non-convenient language for large data analysis. </p>
<p>Solving this problems will take R to other dimension : </p>
<h3>EFFICIENT LARGE DATA ANALYSIS</h3>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="memory1" style="background:;">
<hgroup>
<h2>THE MEMORY PROBLEM</h2>
</hgroup>
<article data-timings="">
<p><strong>R</strong> is using RAM memory to store computed, computing and temporal data.
RAM overflow errors are common when we use large data in R.</p>
<p>There is not only the limitation of the RAM capacity, but also the limitation of contiguous free RAM blocks: To be able to store a vector in R, we need a contiguous free space in RAM memory for that vector.</p>
<p>And sometimes, while having still some hundreds of megabytes of free memory, we do not have any <em>contiguous free block</em> bigger that a few megabytes. </p>
<p>Then, any large vector will not be able to fit into memory and we will get an error.</p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-8" style="background:;">
<article data-timings="">
<p><img src="assets/img/ramf.png" alt="RAM" height="600" width="800" align="middle"> </p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-9" style="background:;">
<hgroup>
<h2>MEMORY R EXAMPLE</h2>
</hgroup>
<article data-timings="">
<pre><code class="r">pr <- function(x){object.size(x) }
xint <- 0L;
N <- 1e6
x <- rep(23L,N); m.int <- pr(x)
x <- rep(0.23,N); m.float <- pr(x)
x <- as.Date("2000-01-01") + 1:N;
m.date <- pr(x)
x <- as.factor(1:N); m.factor <- pr(x)
x <- as.factor(rep(23L,N)); m.factor.2 <- pr(x) # it depends on the content !!!
x <- as.character(1:N); m.char <- pr(x)
x <- as.character(rep(23L,N)); m.char2 <- pr(x) # it depends on the content !!!!
memos <- c(m.int,m.float,m.date,m.char,m.factor)
md <- data.frame(class = c("integer","numeric", "date","character", "factor"), size = round(memos/1e6,digits =1) )
</code></pre>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-10" style="background:;">
<hgroup>
<h2>MEMORY R EXAMPLE (2)</h2>
</hgroup>
<article data-timings="">
<p>So we have this sizes in Mb for vectors of 1 million item:</p>
<pre><code class="r">md
</code></pre>
<pre><code>## class size
## 1 integer 4
## 2 numeric 8
## 3 date 8
## 4 character 36
## 5 factor 40
</code></pre>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-11" style="background:;">
<hgroup>
<h2>MEMORY R EXAMPLE (3)</h2>
</hgroup>
<article data-timings="">
<p>The problem arises when we try to load a vector of large size </p>
<h4>(2e8 integer, which will be 20*3.8=762 Mb) :</h4>
<pre><code class="r">N <- 2e+08
x <- rep(0L, N)
print(object.size(x), units = "Mb")
gc()
</code></pre>
<p>Getting the error: </p>
<pre><code>Error: cannot allocate vector of size 762.9 Mb
</code></pre>
<p>We are unable to load a 762 Mb vector, having 4 Gb of RAM , because we do not have a contiguous free memory block.
We need a way to store efficiently data and temporal computation data into the hard drive, and load dinamically... </p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-12" style="background:;">
<hgroup>
<h2>THE SPEED PROBLEM</h2>
</hgroup>
<article data-timings="">
<p>Dealing with large data needs efficient code to speed up computations.
Non efficient code can result in extremely slow processes, that can last hours or days instead of seconds or minutes. </p>
<p>Special care must be done to produce efficient code, using native R code or some packages and/or other languages.</p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-13" style="background:;">
<hgroup>
<h2>SOLVING ISSUES : MEMORY</h2>
</hgroup>
<article data-timings="">
<p>To solve memory problems we need to write data into disk, instead of loading into RAM memory. </p>
<p>The package <code>ff</code> and <code>ffbase</code> allow to do this and still do very efficient code. </p>
<p>Some main functions of ffbase were <strong>modified or re-coded</strong> to allow more efficiency or avoid memory problems,
that still exist using the ffbase package. </p>
<p>Some internal coding of the package was modified by the authors after we had some useful discussions with them. </p>
<p>The result is that to solve memory problems we use ffbase + some new functions, that avoid almost all the possible memory problems in R, while still keeping efficient code. </p>
<h2>FFBASE + CUSTOM FUNCTIONS --> NO MEMORY ERRORS</h2>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-14" style="background:;">
<hgroup>
<h2>SOLVING ISSUES : SPEED</h2>
</hgroup>
<article data-timings="">
<p>Large data --> speeds factors of 10 , 100 , 1000 or more between efficient and non-efficient code.
To accomplish a speed-optimized code, we have used 3 main procedures : </p>
<ol class = "build incremental">
<li><p><strong>Efficient R code</strong>.
Using vectorized functions, efficient built-in functions, etc.
This will make the code several orders of magnitude faster than other codes (see details in the tutorial files) </p></li>
<li><p><strong>Use data.table</strong> package for "group by" and split analysis.
data.table is a very efficient package that allows to compute "group by", splits, etc.,
analysis in a very efficient way.
We use it for almost any "group by" analysis and for other specific tasks. </p></li>
<li><p><strong>Rcpp</strong>:
C++ code in R.
C++ is one of the fastest languages that we can code in computers. The package "Rcpp" and "inline" ,
allow to create C++ code in R in an "easy" way.
For some tasks, Rcpp code is thousands times faster than R code. We have developed some Rcpp functions to work with large data.
We have also made some tutorials about how, when, and why use Rcpp into R code.</p></li>
</ol>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-15">
<hgroup>
</hgroup>
<article class = 'flexbox vcenter'>
<p>##</p>
<h2>ffbase + efficient R + data.table +</h2>
<h2>Rcpp + custom functions</h2>
<h2>=</h2>
<h2>SPEED-OPTIMIZED CODE</h2>
<p><img src="assets/img/dog.jpg" alt="RAM" height="200" width="300" align="middle"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-16" style="background:;">
<hgroup>
<h2>SOLVING THE PROBLEM : THE INTERNAL PROCESS</h2>
</hgroup>
<article data-timings="">
<h4>All those speed-efficient techniques are combined with ffbase to compute tasks into blocks that can fit into RAM memory, write the result in disk, and proceed with the next block of data loaded into RAM. In this way, we obtain the benefits of speed, avoiding memory problems.</h4>
<p><img src="assets/img/ramff.png" alt="RAMFF" height="600" width="800" align="middle"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-17" style="background:#EDE0CF;">
<hgroup>
<h2>CHOOSING A COMPUTER TO RUN R</h2>
</hgroup>
<article data-timings="">
<p>Due to this memory problem in R, a good computer to work with R would be one with a good RAM memory, working in 64bits OS and having fast write/read hard disk capabilities. </p>
<p>The ideal computer would be : </p>
<ol>
<li>>= 16 GB RAM </li>
<li>SD (solid state) hard disk</li>
<li>64 bits Linux OS // 64bits Mac OSX<br></li>
<li>Fast cpu, etc.</li>
</ol>
<h4>The work we present here is done while working in a <strong>Windows 32bits computer with 4gb RAM</strong>, which is almost the "worst" computer we can get nowadays to work with R.</h4>
<h4>Having this computer, we could work in a very efficient way with databases containing dozens of millions records and a few dozens of columns.</h4>
<p>Using the proposed techniques and functions, <strong>any computer will be able to deal with large data in R</strong>. </p>
<p>And this opens a new possibility for the use of R with large data, for example for <em>less rich countries or for the common desktop computers</em>. </p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-18" style="background:;">
<hgroup>
<h2>LARGE DATA EXAMPLE</h2>
</hgroup>
<article data-timings="">
<p>We compute the same task following different methods and finding their computational duration. </p>
<ol>
<li><p>Task 1:
in a large database, label records whith<br>
date_min <= date < date_max and diagnose == "a" or "b" or "c" as "positive" and others as "negative". </p></li>
<li><p>Task 2:
In a large database, for each ident, find the number of cases with diagnoses "d" , "e" or "f".
If n == 0 label case as "none", if 0< n <= 10 label = few , others: label = many. </p></li>
</ol>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-19" style="background:;">
<hgroup>
<h2>The database:</h2>
</hgroup>
<article data-timings="">
<p>The data (20 million rows) :: </p>
<pre><code>df[1:3,]
ident date diagnose region
324958 1991-07-25 l 070
1339198 1974-03-24 z 065
553078 1978-01-06 s 065
</code></pre>
<p>Structure of the Data :: </p>
<pre><code>str(df[1,])
'data.frame': 1 obs. of 4 variables:
$ ident : Factor w/ 1999919 levels "1000","1001",..: 323942
$ date : Date, format: "1991-07-25"
$ diagnose: Factor w/ 26 levels "a","b","c","d",..: 12
$ region : Factor w/ 18 levels "000","013","014",..: 15
</code></pre>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-20" style="background:;">
<hgroup>
<h2>TASK 1</h2>
</hgroup>
<article data-timings="">
<p>Label records whith date.min <= date < date.max and diagnose == "a" or "b" or "c" as "positive" and others as "negative"</p>
<ul>
<li>A. loop over the rows of the database. </li>
<li>B. mapply </li>
<li>C. vectorized function</li>
<li>D. data.table</li>
<li>E. Rcpp</li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-21" style="background:;">
<hgroup>
<h2>Speed comparison data :</h2>
</hgroup>
<article data-timings="">
<table><thead>
<tr>
<th>method</th>
<th>time</th>
<th>time.relative</th>
<th>time.20e6</th>
<th>label</th>
</tr>
</thead><tbody>
<tr>
<td>loop rows</td>
<td>65.2800</td>
<td>7865</td>
<td>13056.00</td>
<td>3.62 hours</td>
</tr>
<tr>
<td>mapply</td>
<td>65.7000</td>
<td>7916</td>
<td>13140.00</td>
<td>3.65 hours</td>
</tr>
<tr>
<td>apply</td>
<td>27.6400</td>
<td>3330</td>
<td>5528.00</td>
<td>1.5 hours</td>
</tr>
<tr>
<td>vectorized</td>
<td>0.0600</td>
<td>7</td>
<td>12.00</td>
<td>12 seconds</td>
</tr>
<tr>
<td>data.table</td>
<td>0.0500</td>
<td>6</td>
<td>10.00</td>
<td>10 seconds</td>
</tr>
<tr>
<td>Rcpp</td>
<td>0.0083</td>
<td>1</td>
<td>1.66</td>
<td>1.7 seconds</td>
</tr>
</tbody></table>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-22" style="background:;">
<hgroup>
<h2>Graph the results ::</h2>
</hgroup>
<article data-timings="">
<pre><code class="r">plot <- ggplot(speed, aes(x = method, y = time_relative, fill = method, group = 1))
plot <- plot + geom_bar(stat = "identity") + xlab("computing methods") + ylab("relative computing time") +
ggtitle("Relative computing times for different methods in R \n over a 20 million row data frame ")
plot2 <- plot + geom_text(data = speed, aes(x = method, y = time_relative, label = time_20e6),
vjust = -0.5, size = 4) + theme(plot.title = element_text(size = 16, vjust = 1),
legend.position = c(0.82, 0.67), axis.title.x = element_text(vjust = -1,
size = rel(1.2)), axis.title.y = element_text(angle = 90, vjust = 0,
size = rel(1.2)), plot.margin = unit(c(1, 1, 1, 1), "lines"))
</code></pre>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-23" style="background:;">
<hgroup>
<h2>Speed results</h2>
</hgroup>
<article data-timings="">
<p><img src="assets/img/speed.jpg" alt="Speed" height="600" width="800" align="middle"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-24" style="background:;">
<hgroup>
<h2>TASK 2</h2>
</hgroup>
<article data-timings="">
<p>In a large database, for each ident, find the number of cases with diagnoses "d" , "e" or "f". </p>
<p>If n == 0 label case as "none", if 0< n <= 6 label = few , others: label = many. </p>
<p>As a general rule, timings for group_by computations will depend on the number of levels for the variable grouped.</p>
<p>For same number of levels, the fact that data is ordered or not is not very important for some methods. </p>
<p>For other methods, the ordering can improve the speed (as for ffdfdply for example).</p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-25" style="background:;">
<hgroup>
<h2>Speed comparison data :</h2>
</hgroup>
<article data-timings="">
<p>(2 million rows)</p>
<table><thead>
<tr>
<th>method</th>
<th>time.relative</th>
<th>label 2e6</th>
<th>label 20e6</th>
</tr>
</thead><tbody>
<tr>
<td>tapply</td>
<td>3992</td>
<td>27 minutes</td>
<td>4.1 hours</td>
</tr>
<tr>
<td>aggregate</td>
<td>4212</td>
<td>26 minutes</td>
<td>4.3 hours</td>
</tr>
<tr>
<td>data.table</td>
<td>20</td>
<td>7 seconds</td>
<td>1.24 minutes</td>
</tr>
<tr>
<td>sqldf</td>
<td>17</td>
<td>6.3 seconds</td>
<td>1.05 minutes</td>
</tr>
<tr>
<td>Rcpp</td>
<td>1</td>
<td>0.37 seconds</td>
<td>3.7 seconds</td>
</tr>
</tbody></table>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-26" style="background:;">
<hgroup>
<h2>Speed results</h2>
</hgroup>
<article data-timings="">
<p><img src="assets/img/speed2.jpg" alt="Speed" height="600" width="800" align="middle"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-27" style="background:;">
<hgroup>
<h2>DOCUMENTATION : REPIDEMIOLOGY WEBSITE</h2>
</hgroup>
<article data-timings="">
<p><a href="http://repidemiology.wordpress.com/">http://repidemiology.wordpress.com/</a> , contains tutorials, videos and code about R.
It is meant to be a site where beginners and advanced can learn and work with large data. </p>
<p><img src="assets/img/press.jpg" alt="Speed" height="600" width="800" align="middle"></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-28" style="background:;">
<hgroup>
<h2>SOLVING THE PROBLEM : NEW FUNCTIONS</h2>
</hgroup>
<article data-timings="">
<h2>( SUMMARY LIST)</h2>
<ul>
<li>order_ffdf (data=,order_cols=,splits=,verbose=)</li>
<li>subset_ffdf (data=,condition_cols=,condition=,splits=,verbose=)</li>
<li>cbind_ffdf (x=,y=)</li>
<li>append_<em>ffdf (x=,y</em>=,chunk_s=)</li>
<li>merge_by (x=,y=,keycols=,split_name=,chk=)</li>
<li>index_ffdf (x=,idx=,verbose=)</li>
<li>index_ffdf_split (x=,idx=,nsplits=,verbose=)</li>
<li>split_<em>apply_cpp (input_data=,split_vector=,chk_size</em>=,fu=)</li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-29" style="background:;">
<hgroup>
<h2>SOME NEW UTILITIES</h2>
</hgroup>
<article data-timings="">
<ul>
<li>redir_ffdf (ffdf=,newdir= ) </li>
<li>redir_ffdf_ffdfsave (ff=,newdir=)<br></li>
<li>is_ordered_by_row_col (data=,col_name=) </li>
<li>quote_names()<br></li>
<li>qn()<br></li>
<li>lagpad (x=)</li>
<li>asDate (x=)</li>
<li>txt.perc (x=,dec=)</li>
<li>fill.vec (x=,fill="*")</li>
<li>fill.vec.all.same (x=,fill="*") </li>
<li>fill.data.frame (x=,fill.char="*",make.factors=FALSE)</li>
<li>fill.data.frame.all.same (df,"_")<br></li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-30" style="background:;">
<hgroup>
<h2>APPENDIX</h2>
</hgroup>
<article data-timings="">
<p>This documents was made using ONLY R.
WIth packages <code>shiny</code> and <code>slidify</code></p>
<p>This is an example of the code used : </p>
<pre><code>>1. OPEN SOURCE
>2. it is used in an increasing number of research centers and media.
>3. with some patience, you can do almost ANYTHING with your data.
>4. is similar to other high performance languages, like C++ , python, etc.
```r
pr <- function(x){object.size(x) }
xint <- 0L;
```
</code></pre>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-31" style="background:;">
<hgroup>
<h2>APPENDIX 2 : INTERACTIVITY</h2>
</hgroup>
<article data-timings="">
<p>Counts of diagnoses = a,b,c for different regions. </p>
<div id='chart1' class='rChart nvd3'></div>
<script type='text/javascript'>
$(document).ready(function(){
drawchart1()
});
function drawchart1(){
var opts = {
"dom": "chart1",
"width": 800,
"height": 400,
"x": "diagnoseklasse",
"y": "Freq",
"group": "screening_region2007",
"type": "multiBarChart",
"id": "chart1"
},
data = [
{
"screening_region2007": "X",
"diagnoseklasse": "X",
"Freq": 408437
},
{
"screening_region2007": "X",
"diagnoseklasse": "c",
"Freq": 362670
},
{
"screening_region2007": "X",
"diagnoseklasse": "d",
"Freq": 15450
},
{
"screening_region2007": "X",
"diagnoseklasse": "a",
"Freq": 6588
},
{
"screening_region2007": "C",
"diagnoseklasse": "X",
"Freq": 23277
},
{
"screening_region2007": "A",
"diagnoseklasse": "X",
"Freq": 36215
},
{
"screening_region2007": "B",
"diagnoseklasse": "X",
"Freq": 15856
},
{
"screening_region2007": "E",
"diagnoseklasse": "X",
"Freq": 10961
},
{
"screening_region2007": "D",
"diagnoseklasse": "X",
"Freq": 23849
},
{
"screening_region2007": "X",
"diagnoseklasse": "b",
"Freq": 6038
},
{
"screening_region2007": "C",
"diagnoseklasse": "c",
"Freq": 3202
},
{
"screening_region2007": "D",
"diagnoseklasse": "d",
"Freq": 177
},
{
"screening_region2007": "E",
"diagnoseklasse": "c",
"Freq": 1685
},
{
"screening_region2007": "A",
"diagnoseklasse": "c",
"Freq": 6103
},
{
"screening_region2007": "D",
"diagnoseklasse": "c",
"Freq": 3330
},
{
"screening_region2007": "B",
"diagnoseklasse": "d",
"Freq": 146
},
{
"screening_region2007": "B",
"diagnoseklasse": "c",
"Freq": 2004
},
{
"screening_region2007": "D",
"diagnoseklasse": "a",
"Freq": 81
},
{
"screening_region2007": "E",
"diagnoseklasse": "d",
"Freq": 32
},
{
"screening_region2007": "C",
"diagnoseklasse": "a",
"Freq": 61
},
{
"screening_region2007": "A",
"diagnoseklasse": "d",
"Freq": 214
},
{
"screening_region2007": "E",
"diagnoseklasse": "a",
"Freq": 36
},
{
"screening_region2007": "C",
"diagnoseklasse": "d",
"Freq": 103
},
{
"screening_region2007": "A",
"diagnoseklasse": "a",
"Freq": 122
},
{
"screening_region2007": "D",
"diagnoseklasse": "b",
"Freq": 83
},
{
"screening_region2007": "B",
"diagnoseklasse": "a",
"Freq": 43
},
{
"screening_region2007": "C",
"diagnoseklasse": "b",
"Freq": 78
},
{
"screening_region2007": "E",
"diagnoseklasse": "b",
"Freq": 37
},
{
"screening_region2007": "A",
"diagnoseklasse": "b",
"Freq": 50
},
{
"screening_region2007": "B",
"diagnoseklasse": "b",
"Freq": 17
}
]
var data = d3.nest()
.key(function(d){
return opts.group === undefined ? 'main' : d[opts.group]
})
.entries(data)
nv.addGraph(function() {
var chart = nv.models[opts.type]()
.x(function(d) { return d[opts.x] })
.y(function(d) { return d[opts.y] })
.width(opts.width)
.height(opts.height)
d3.select("#" + opts.id)
.append('svg')
.datum(data)
.transition().duration(500)
.call(chart);
nv.utils.windowResize(chart.update);
return chart;
});
};
</script>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="backdrop"></slide>
</slides>
<div class="pagination pagination-small" id='io2012-ptoc' style="display:none;">
<ul>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=1 title='R'>
1
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=2 title='Why R ??'>
2
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=3 title='Why R?'>
3
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=4 title='SAS VS R'>
4
</a>