/
slides.html
876 lines (671 loc) · 25.2 KB
/
slides.html
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
<!DOCTYPE html>
<html xmlns="http://www.w3.org/1999/xhtml" lang="" xml:lang="">
<head>
<title>Human vs computer</title>
<meta charset="utf-8" />
<meta name="author" content="Professor Di Cook Monash University" />
<link href="libs/font-awesome-animation/font-awesome-animation-emi.css" rel="stylesheet" />
<script src="libs/fontawesome/js/fontawesome-all.min.js"></script>
<script src="libs/kePrint/kePrint.js"></script>
<link href="libs/font-awesome/css/fontawesome-all.min.css" rel="stylesheet" />
<link rel="stylesheet" href="xaringan-themer.css" type="text/css" />
<link rel="stylesheet" href="mytheme.css" type="text/css" />
</head>
<body>
<textarea id="source">
class: inverse middle
background-image: url(images/redgum.jpg)
background-position: 110% 50%
background-size: 50%
# Human vs computer
## .pink[*In Visualising Data, Who Wins?*]
## Experiments with building computer vision models for reading residual plots
### .lightgreen[Professor Di Cook <br> Monash University]
.green[DSSV '19, Kyoto <br>Aug 14, 2019]
Slides can be viewed at [http://dicook.org/files/dssv19/slides#1](http://dicook.org/files/dssv19/slides#1)
---
class: inverse middle
<span><i class="fas fa-hand-paper fa-2x faa-flash animated-hover faa-slow " style=" color:#FD4D6A;"></i></span> <span class="fa-2x faa-flash animated-hover faa-slow " style=" color:#959B4C; display: -moz-inline-stack; display: inline-block; transform: rotate(0deg);">Hands up if you have fitted a model in the last week</span>
--
<br>
<br>
<br>
<span class="fa-2x faa-flash animated-hover faa-slow " style=" color:#D8BA69; display: -moz-inline-stack; display: inline-block; transform: rotate(0deg);">Keep your hand up if you also looked at a residual plot from that model fit</span>
--
---
background-image: url(images/cartoon.png)
background-size: contain
---
background-image: url(images/housekeeping.png)
background-size: contain
---
background-image: url(images/robot.png)
background-size: contain
---
# Outline
- Motivation: inspiration for these experiments
- Methods
- Deep learning: image recognition
- Visual inference: process and framework
- Experiments
- Design
- Results
- Future
- Acknowledgements, resources and references
---
background-image: url(images/simchoni.png)
background-size: 80%
# True motivation is Giora's blog post
---
.left-code[
Trained a computer vision two ways:
- classification, significant correlation vs not
- regression, to predict the correlation
.green[Success, picked plot 16]
]
.right-plot[
<img src="images/simchoni_test1.png" width="100%" />
]
---
.left-code[
Trained a computer vision two ways:
- classification, significant correlation vs not
- regression, to predict the correlation
.green[Success, failed to pick plot 4]
]
.right-plot[
<img src="images/simchoni_test2.png" width="100%" />
]
---
.left-code[
Trained a computer vision two ways:
- classification, significant correlation vs not
- regression, to predict the correlation
.pink[Fail! Doesn't see the strong nonlinear association. Picks the most linear.]
]
.right-plot[
<img src="images/simchoni_test3.png" width="100%" />
]
---
.left-column[
# Deep learning
]
.right-column[
<img src="images/image_classification.png" width="100%" />
.footnote[Source: [Abdellatif Abdelfattah](https://medium.com/@tifa2up/image-classification-using-deep-neural-networks-a-beginner-friendly-approach-using-tensorflow-94b0a090ccd4)]
]
---
.left-column[
# Deep learning
]
.right-column[
<img src="https://miro.medium.com/max/1230/1*DGDcAQmm0e0kW_1iON1e-Q.jpeg" width="70%" />
.footnote[Source: [Chihuahua or Muffin? Brad Folkens](https://blog.cloudsight.ai/chihuahua-or-muffin-1bdf02ec1680)]
]
---
.left-column[
# Deep learning
### Neural network
]
.right-column[
A linear model can be thought of as a neural network:
<img src="images/linear_model_nn.png" width="100%" />
.footnote[Source: [Cheng and Titterington (1994)](https://projecteuclid.org/euclid.ss/1177010638)]
]
---
.left-column[
# Deep learning
### Neural network
]
.right-column[
A simple neural network, contains a hidden layer, to increase/decrease dimensionality, and a transformation through a (logistic) function:
<img src="images/hidden_layer_nn.png" width="100%" />
]
---
.left-column[
# Deep learning
### Neural network
]
.right-column[
<img src="images/nnet-hidden-fill.png" width="100%" />
<img src="images/nnet-best-fill.png" width="40%" />
.footnote[Example from [Wickham et al (2015) "Removing the blindfold"](https://onlinelibrary.wiley.com/doi/abs/10.1002/sam.11271)]
]
---
.left-column[
# Fitting neural networks is difficult
<img src="images/nnet-best-fill.png" width="100%" />
.footnote[Example from [Wickham et al (2015) "Removing the blindfold"](https://onlinelibrary.wiley.com/doi/abs/10.1002/sam.11271)]
]
.right-plot[
<img src="images/nnet-value-accuracy.png" width="100%" />
<img src="images/nnet-all.png" width="80%" />
]
---
.left-column[
# Deep learning
### Neural network
### Convolutional NN
]
.right-plot[
Thinking about images as data, requires unpacking into variables:
<img src="images/img_nn.jpg" width="80%" />
]
---
.left-column[
# Deep learning
### Neural network
### Convolutional NN
]
.right-column[
- Proximity in an image matters though.
- The dimension reduction takes into account of neighbouring cells in the image, using filters
<img src="images/covnet2.jpg" width="100%" />
]
---
.left-column[
# Deep learning
### Neural network
### Convolutional NN
]
.right-column[
Applying multiple filters
<img src="images/covnet1.jpg" width="100%" />
]
---
.left-column[
# Deep learning
### Neural network
### Convolutional NN
### Software
]
.right-column[
The [keras](https://keras.rstudio.com) package in R, interfaces to python software, provides a convenient way to fit the convolutional neural network.
<img src="images/dlwR.png" width="40%" />
]
---
class: center
background-image: url(images/robot.png)
background-size: 100%
## Training the robot
---
# Model set up overview
- Formulate the problem as a classification task: "good" vs "bad" residual plot
- Simulate an enormous number of both types of residual plots
- Train the model on half of these
- Choose the best model by accuracy with the other half
<img src="images/plot_turk2_300_70_5_2_7.png" width="19%" /> <img src="images/plot_turk2_300_70_5_2_1.png" width="19%" />
.footnote[beta=0.7, sigma=12, n=300]
---
background-image: url(images/training.png)
background-size: 100%
.footnote[Source: Shuofan Zhang]
---
# Model training
1. Simulate data (X, Y) from the null and the alternative models.
2. Generate scatterplots of X and Y
3. Save scatter plots as 150 × 150 pixels images
4. Train a deep learning classifier to recognise the patterns from two groups
5. Test the model's performance on new data and compute the accuracy
6. One model chosen as the best, to compare performance with humans
---
class: center middle
background-image: url(images/housekeeping.png)
background-size: 150%
## How the human evaluated residuals plots
---
.left-column[
# Human evaluation
### Lineup protocol
]
.right-column[
- Statistical inference with graphics, follows classical hypothesis testing procedures
- Lineup protocol embeds the data plot among a field of null plots (comparison of test statistic with sampling distribution)
- Human observers pick plot that is most different from the others
- `\(p\)`-value calculated from the probability that the data plot is selected by chance
[Buja et al (2009)](http://rsta.royalsocietypublishing.org/content/367/1906/4361) ideas, [Majumder et al (2013)](https://www.tandfonline.com/doi/abs/10.1080/01621459.2013.808157) validation
]
---
background-image: url(images/inference_diagram.png)
background-size: contain
.footnote[[Majumder et al (2013)](https://www.tandfonline.com/doi/abs/10.1080/01621459.2013.808157)]
---
.left-column[
# Human evaluation
### Lineup protocol
### Database
]
.right-column[
- [Majumder et al (2013)](https://www.tandfonline.com/doi/abs/10.1080/01621459.2013.808157) conducted validation study to compare the performance of the lineup protocol, assessed by human evaluators, in comparison to the classical test
- Experiment 2 examined `\(H_o: \beta_k = 0 ~~vs ~~H_a: \beta_k \neq 0\)`
- 70 lineups of size 20 plots: `\(n=100, 300; \beta \in [-6, 4.5]; \sigma=5, 12\)`
- 351 evaluations by human subjects
- Amazon's Mechanical Turk used to recruit observers
]
---
.left-column[
# Human evaluation
### Lineup protocol
### Database
### Try it
]
.right-column[
Let's do a couple:
<br>
<br>
*From the field of 20 plots shown, pick the plot with the strongest linear relationship between the two variables.*
]
---
background-image: url(images/plot_turk2_300_150_5_3.png)
background-size: contain
---
.left-column[
# Human evaluation
## Lineup protocol
## Database
## Try it
]
.right-column[
<br>
<br>
<br>
*If you chose 19 you picked the data plot.*
In the Turk study 76/78 observers chose plot 19. The estimated p-value is 0. It is very unlikely to see this result if the data plot was a sample a population with no relationahip between the variables.
]
---
background-image: url(images/plot_turk2_300_10_5_3.png)
background-size: contain
---
.left-column[
# Human evaluation
### Lineup protocol
### Database
### Try it
]
.right-column[
<br>
<br>
<br>
*If you chose 18 you picked the data plot.*
In the Turk study 0/19 observers chose plot 18. The estimated p-value is 1. It is very likely to see this result if the data plot was a sample a population with no relationahip between the variables.
11/19 chose plot number 19. This plot is the plot in the field that has the smallest p-value. People do tend to pick the plot that has the smallest p-value (or the most structure).
]
---
.left-column[
# Human evaluation
### Lineup protocol
### Database
### Try it
### Validation
.footnote[Source: [Majumder et al (2013)](https://www.tandfonline.com/doi/abs/10.1080/01621459.2013.808157)]
]
.right-plot[
Effect `\(= \frac{\sqrt{n}\times |\beta|}{\sigma}\)`
<img src="images/majumder1.png" width="70%" />
<img src="images/majumder2.png" width="30%" />
]
---
background-image: url(images/diaghm.png)
background-size: 100%
.footnote[Source: Shuofan Zhang]
---
.left-column[
# Human vs computer
### Comparison
]
.right-plot[
`\(H_o: \beta_k = 0 ~~vs ~~H_a: \beta_k \neq 0\)`
Linear vs no relationship (null)
]
---
.left-column[
# Human vs computer
### Comparison
### Computer model
]
.pull-right[
Simulation
`$$Y_i = \beta_0 + \beta_1 X_{i} + \varepsilon_i, ~~i=1, \dots , n$$`
- `\(X \sim N[0,\ 1]\)`
- `\(\beta_0 = 0\)` (intercept)
- `\(\beta_1\sim U[-10, -0.1] \bigcup [0.1, 10]\)` (linear, null when `\(\beta_1=0\)`)
- `\(\varepsilon\sim N(0, \sigma^2) \ where\ \sigma \sim U[1,12]\)`
- `\(n=U[50,500]\)`
- 200,000 from each linear and null scenario generated
<img src="images/covnet3.png" style="width: 50%; align: center" />
]
---
.left-column[
# Human vs computer
### Comparison
### Computer model
### Testing
]
.pull-right[
### Computer model prediction
- Re-generate the 70 *data plots* using the same data in Turk study (without null plots)
- Use the computer model to predict whether the 70 data plots were "linear" or "null"
- The computer model's predicted accuracy over the 70 data plots are recorded as the model's performance.
]
---
.left-column[
# Human vs computer
### Comparison
### Computer model
### Testing
]
.pull-right[
### Human subjects results
- Calculate `\(p\)`-value associated with each lineup using the binomial formula (from Majumder), with `\(N\)`=number of evaluations and k=number of people choosing data plot
- Draw conclusion: reject the null when the calculated `\(p\)`-value is smaller than `\(\alpha\)`.
- The accuracy of the conclusions over the 70 lineups
]
---
class: inverse middle center
<span><i class="fas fa-hourglass fa-2x faa-tada animated faa-slow " style=" color:#FD4D6A;"></i></span>
### Stay tuned for the result!
--
## Experiment repeated
---
# Repeat of experiment
Using same sample of `\(n\)`, `\(\beta\)`, `\(\sigma\)`, new data generated, and images created numerically by binning (to 30x30 pixels), counting and scaling counts to 0-255.
Keras model fitted with 60,000 training images for each class, linear and not.
Accuracy with simulated test data, 93%. Null error 0.0179, linear error 0.1176
.footnote[Code available in the file `keras_correlation.r`]
--
.pink[Its blindingly fast!]
---
background-image: url(images/drumroll.gif)
background-size: 100%
---
.left-column[
# Human vs computer
### Comparison
### Computer model
### Testing
### Results
]
.right-plot[
# Results 1
<img src="images/results.png" width="100%" />
.pink[Humans beat computers.]
]
---
.left-column[
# Human vs computer
### Comparison
### Computer model
### Testing
### Results
]
.right-plot[
# Results 2
<img src="images/new_computer_v_human.png" width="80%" />
.pink[Humans beat computers, and match the power of the theoretical t-test (uniformly most powerful test).]
]
---
# Comparison of human and computer.
<table>
<tr> <th> </th> <th> </th> <th colspan="2"> Computer </th> </tr>
<tr> <th> </th> <th> </th> <th> Not </th> <th> Linear </th> </th> </tr>
<tr> <th rowspan="2"> Human </th> <th> Not </th> <td> 27 </td> <td> 0 </td> </tr>
<tr> <th> Linear </th> <td> 15 </td> <td> 28 </td> </tr>
</table>
<br>
<br>
.pink[Computer tends to predict too many as not linear.]
---
background-image: url(images/simchoni2.png)
background-size: contain
[Giora Simchoni's JSM 2019 talk](http://giorasimchoni.com/deep_visual_inference.html#/1)
---
# Summary
- Computer model trained specifically on detecting linear vs no relationship, comes close but does not beat human evaluation
- Results are consistent: images vs binned data, image size, number of training samples
---
.left-code[
# Discussion
Graphics are important in data analysis
<img src="slides_files/figure-html/unnamed-chunk-3-1.png" width="80%" />
.footnote[Source: [Matejka, J. and Fitzmaurice, G. (2017)](https://www.autodeskresearch.com/publications/samestats) and [Locke, S. and D'Agostino McGowan, L. (2018)](https://CRAN.R-project.org/package=datasauRus)]
]
.right-plot[
Same numerical statistics but radically different structure
<table class="table" style="margin-left: auto; margin-right: auto;">
<thead>
<tr>
<th style="text-align:left;"> dataset </th>
<th style="text-align:right;"> mx </th>
<th style="text-align:right;"> my </th>
<th style="text-align:right;"> sx </th>
<th style="text-align:right;"> sy </th>
<th style="text-align:right;"> r </th>
</tr>
</thead>
<tbody>
<tr>
<td style="text-align:left;"> away </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> bullseye </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> circle </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> dino </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> dots </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> h_lines </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> high_lines </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> slant_down </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> star </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> v_lines </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> wide_lines </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
<tr>
<td style="text-align:left;"> x_shape </td>
<td style="text-align:right;"> 54.3 </td>
<td style="text-align:right;"> 47.8 </td>
<td style="text-align:right;"> 16.8 </td>
<td style="text-align:right;"> 26.9 </td>
<td style="text-align:right;"> -0.1 </td>
</tr>
</tbody>
</table>
]
---
.left-code[
[Residual (Sur)Realism](https://www4.stat.ncsu.edu/~stefanski/nsf_supported/hidden_images/stat_res_plots.html#Download_Data_Sets) or how to hide images in your residuals.
<img src="slides_files/figure-html/unnamed-chunk-5-1.png" width="504" />
.footnote[Looks boring]
]
.right-plot[
Fit a linear model, plot the residuals
<img src="slides_files/figure-html/unnamed-chunk-6-1.png" width="504" />
]
---
# To the future <i class="fas fa-rocket "></i>
- The lineup protocol, from visual inference, provides a way to check and validate the models
- Need to build up more human subject evaluations
- Model checking may be systematic enough to automate.
- Computer only needs to relieve us of most of the work, after all.
- Expand the computer model with simulations of all sorts of departures from no relationship
- Web service for uploading plots?
---
background-image: url(images/Rosie.png)
background-size: contain
---
class: inverse
# Joint work!
- Deep learning model, human vs computer comparison: Monash Masters thesis by Shuofan Zhang (inspired by a blog post by [Giora Simchoni](http://giorasimchoni.com/2018/02/07/2018-02-07-book-em-danno/))
- Inference: Heike Hofmann, Mahbub Majumder, Andreas Buja, Hadley Wickham, Eric Hare, Susan Vanderplas, Adam Loy, Niladri Roy Chowdhury, Nat Tomasetti.
Contact: [<i class="fas fa-envelope "></i>](http://www.dicook.org) dicook@monash.edu [<i class="fab fa-twitter "></i>](https://twitter.com/visnut) visnut [<i class="fab fa-github "></i>](https://github.com/dicook) dicook
.footnote[Slides made with Rmarkdown, xaringan package by Yihui Xie, modified xaringanthemer by Garrick Aden-Buie, and the anicon package by Emi Tanaka, with icon package by Mitch O'Hara-Wild. Available at [https://github.com/dicook/DSSV19](https://github.com/dicook/files/DSSV19).]
---
# Further reading
- Buja et al (2009) Statistical Inference for Exploratory Data Analysis and Model Diagnostics, Roy. Soc. Ph. Tr., A
- Majumder et al (2013) Validation of Visual Statistical Inference, Applied to Linear Models, JASA
- Wickham et al (2010) Graphical Inference for Infovis, InfoVis, Best paper
- Hofmann et al (2012) Graphical Tests for Power Comparison of Competing Design, InfoVis
- Loy et al (2017) Model Choice and Diagnostics for Linear Mixed-Effects Models Using Statistics on Street Corners, JCGS
---
class: middle center
<a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/"><img alt="Creative Commons License" style="border-width:0" src="https://i.creativecommons.org/l/by-sa/4.0/88x31.png" /></a><br />This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.
</textarea>
<style data-target="print-only">@media screen {.remark-slide-container{display:block;}.remark-slide-scaler{box-shadow:none;}}</style>
<script src="https://remarkjs.com/downloads/remark-latest.min.js"></script>
<script>var slideshow = remark.create({
"ratio": "16:9",
"highlightStyle": "github",
"highlightLines": true,
"countIncrementalSlides": false
});
if (window.HTMLWidgets) slideshow.on('afterShowSlide', function (slide) {
window.dispatchEvent(new Event('resize'));
});
(function(d) {
var s = d.createElement("style"), r = d.querySelector(".remark-slide-scaler");
if (!r) return;
s.type = "text/css"; s.innerHTML = "@page {size: " + r.style.width + " " + r.style.height +"; }";
d.head.appendChild(s);
})(document);
(function(d) {
var el = d.getElementsByClassName("remark-slides-area");
if (!el) return;
var slide, slides = slideshow.getSlides(), els = el[0].children;
for (var i = 1; i < slides.length; i++) {
slide = slides[i];
if (slide.properties.continued === "true" || slide.properties.count === "false") {
els[i - 1].className += ' has-continuation';
}
}
var s = d.createElement("style");
s.type = "text/css"; s.innerHTML = "@media print { .has-continuation { display: none; } }";
d.head.appendChild(s);
})(document);
// delete the temporary CSS (for displaying all slides initially) when the user
// starts to view slides
(function() {
var deleted = false;
slideshow.on('beforeShowSlide', function(slide) {
if (deleted) return;
var sheets = document.styleSheets, node;
for (var i = 0; i < sheets.length; i++) {
node = sheets[i].ownerNode;
if (node.dataset["target"] !== "print-only") continue;
node.parentNode.removeChild(node);
}
deleted = true;
});
})();</script>
<script>
(function() {
var links = document.getElementsByTagName('a');
for (var i = 0; i < links.length; i++) {
if (/^(https?:)?\/\//.test(links[i].getAttribute('href'))) {
links[i].target = '_blank';
}
}
})();
</script>
<script>
slideshow._releaseMath = function(el) {
var i, text, code, codes = el.getElementsByTagName('code');
for (i = 0; i < codes.length;) {
code = codes[i];
if (code.parentNode.tagName !== 'PRE' && code.childElementCount === 0) {
text = code.textContent;
if (/^\\\((.|\s)+\\\)$/.test(text) || /^\\\[(.|\s)+\\\]$/.test(text) ||
/^\$\$(.|\s)+\$\$$/.test(text) ||
/^\\begin\{([^}]+)\}(.|\s)+\\end\{[^}]+\}$/.test(text)) {
code.outerHTML = code.innerHTML; // remove <code></code>
continue;
}
}
i++;
}
};
slideshow._releaseMath(document);
</script>
<!-- dynamically load mathjax for compatibility with self-contained -->
<script>
(function () {
var script = document.createElement('script');
script.type = 'text/javascript';
script.src = 'https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-MML-AM_CHTML';
if (location.protocol !== 'file:' && /^https?:/.test(script.src))
script.src = script.src.replace(/^https?:/, '');
document.getElementsByTagName('head')[0].appendChild(script);
})();
</script>
</body>
</html>