-
Notifications
You must be signed in to change notification settings - Fork 81
/
Copy pathplot_gradient_boosting_quantile.html
729 lines (676 loc) · 75.5 KB
/
plot_gradient_boosting_quantile.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
<!DOCTYPE html>
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
<head>
<meta charset="utf-8">
<meta name="generator" content="Docutils 0.19: https://docutils.sourceforge.io/" />
<meta property="og:title" content="Prediction Intervals for Gradient Boosting Regression" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/auto_examples/ensemble/plot_gradient_boosting_quantile.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="This example shows how quantile regression can be used to create prediction intervals. Generate some data for a synthetic regression problem by applying the function f to uniformly sampled random i..." />
<meta property="og:image" content="https://scikit-learn.org/stable/_static/scikit-learn-logo-small.png" />
<meta property="og:image:alt" content="scikit-learn" />
<meta name="description" content="This example shows how quantile regression can be used to create prediction intervals. Generate some data for a synthetic regression problem by applying the function f to uniformly sampled random i..." />
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Prediction Intervals for Gradient Boosting Regression — scikit-learn 1.4.dev0 documentation</title>
<link rel="canonical" href="http://scikit-learn.org/stable/auto_examples/ensemble/plot_gradient_boosting_quantile.html" />
<link rel="shortcut icon" href="../../_static/favicon.ico"/>
<link rel="stylesheet" href="../../_static/css/vendor/bootstrap.min.css" type="text/css" />
<link rel="stylesheet" href="../../_static/pygments.css" type="text/css" />
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
<link rel="stylesheet" href="../../_static/copybutton.css" type="text/css" />
<link rel="stylesheet" href="../../_static/plot_directive.css" type="text/css" />
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Vibur" type="text/css" />
<link rel="stylesheet" href="../../_static/jupyterlite_sphinx.css" type="text/css" />
<link rel="stylesheet" href="../../_static/sg_gallery.css" type="text/css" />
<link rel="stylesheet" href="../../_static/sg_gallery-binder.css" type="text/css" />
<link rel="stylesheet" href="../../_static/sg_gallery-dataframe.css" type="text/css" />
<link rel="stylesheet" href="../../_static/sg_gallery-rendered-html.css" type="text/css" />
<link rel="stylesheet" href="../../_static/css/theme.css" type="text/css" />
<script id="documentation_options" data-url_root="../../" src="../../_static/documentation_options.js"></script>
<script src="../../_static/js/vendor/jquery-3.6.3.slim.min.js"></script>
</head>
<body>
<nav id="navbar" class="sk-docs-navbar navbar navbar-expand-md navbar-light bg-light py-0">
<div class="container-fluid sk-docs-container px-0">
<a class="navbar-brand py-0" href="../../index.html">
<img
class="sk-brand-img"
src="../../_static/scikit-learn-logo-small.png"
alt="logo"/>
</a>
<button
id="sk-navbar-toggler"
class="navbar-toggler"
type="button"
data-toggle="collapse"
data-target="#navbarSupportedContent"
aria-controls="navbarSupportedContent"
aria-expanded="false"
aria-label="Toggle navigation"
>
<span class="navbar-toggler-icon"></span>
</button>
<div class="sk-navbar-collapse collapse navbar-collapse" id="navbarSupportedContent">
<ul class="navbar-nav mr-auto">
<li class="nav-item">
<a class="sk-nav-link nav-link" href="../../install.html">Install</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="../../user_guide.html">User Guide</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="../../modules/classes.html">API</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" href="../index.html">Examples</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link" target="_blank" rel="noopener noreferrer" href="https://blog.scikit-learn.org/">Community</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../getting_started.html" >Getting Started</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../tutorial/index.html" >Tutorial</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../whats_new/v1.4.html" >What's new</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../glossary.html" >Glossary</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../developers/index.html" >Development</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../faq.html" >FAQ</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../support.html" >Support</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../related_projects.html" >Related packages</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../roadmap.html" >Roadmap</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../governance.html" >Governance</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="../../about.html" >About us</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
</li>
<li class="nav-item">
<a class="sk-nav-link nav-link nav-more-item-mobile-items" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
</li>
<li class="nav-item dropdown nav-more-item-dropdown">
<a class="sk-nav-link nav-link dropdown-toggle" href="#" id="navbarDropdown" role="button" data-toggle="dropdown" aria-haspopup="true" aria-expanded="false">More</a>
<div class="dropdown-menu" aria-labelledby="navbarDropdown">
<a class="sk-nav-dropdown-item dropdown-item" href="../../getting_started.html" >Getting Started</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../tutorial/index.html" >Tutorial</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../whats_new/v1.4.html" >What's new</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../glossary.html" >Glossary</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../developers/index.html" >Development</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../faq.html" >FAQ</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../support.html" >Support</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../related_projects.html" >Related packages</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../roadmap.html" >Roadmap</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../governance.html" >Governance</a>
<a class="sk-nav-dropdown-item dropdown-item" href="../../about.html" >About us</a>
<a class="sk-nav-dropdown-item dropdown-item" href="https://github.com/scikit-learn/scikit-learn" >GitHub</a>
<a class="sk-nav-dropdown-item dropdown-item" href="https://scikit-learn.org/dev/versions.html" >Other Versions and Download</a>
</div>
</li>
</ul>
<div id="searchbox" role="search">
<div class="searchformwrapper">
<form class="search" action="../../search.html" method="get">
<input class="sk-search-text-input" type="text" name="q" aria-labelledby="searchlabel" />
<input class="sk-search-text-btn" type="submit" value="Go" />
</form>
</div>
</div>
</div>
</div>
</nav>
<div class="d-flex" id="sk-doc-wrapper">
<input type="checkbox" name="sk-toggle-checkbox" id="sk-toggle-checkbox">
<label id="sk-sidemenu-toggle" class="sk-btn-toggle-toc btn sk-btn-primary" for="sk-toggle-checkbox">Toggle Menu</label>
<div id="sk-sidebar-wrapper" class="border-right">
<div class="sk-sidebar-toc-wrapper">
<div class="btn-group w-100 mb-2" role="group" aria-label="rellinks">
<a href="plot_forest_iris.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Plot the decision surfaces of ensembles of trees on the iris dataset">Prev</a><a href="index.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Ensemble methods">Up</a>
<a href="plot_bias_variance.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Single estimator versus bagging: bias-variance decomposition">Next</a>
</div>
<div class="alert alert-danger p-1 mb-2" role="alert">
<p class="text-center mb-0">
<strong>scikit-learn 1.4.dev0</strong><br/>
<a href="http://scikit-learn.org/dev/versions.html">Other versions</a>
</p>
</div>
<div class="alert alert-warning p-1 mb-2" role="alert">
<p class="text-center mb-0">
Please <a class="font-weight-bold" href="../../about.html#citing-scikit-learn"><string>cite us</string></a> if you use the software.
</p>
</div>
<div class="sk-sidebar-toc">
<ul>
<li><a class="reference internal" href="#">Prediction Intervals for Gradient Boosting Regression</a><ul>
<li><a class="reference internal" href="#fitting-non-linear-quantile-and-least-squares-regressors">Fitting non-linear quantile and least squares regressors</a></li>
<li><a class="reference internal" href="#analysis-of-the-error-metrics">Analysis of the error metrics</a></li>
<li><a class="reference internal" href="#calibration-of-the-confidence-interval">Calibration of the confidence interval</a></li>
<li><a class="reference internal" href="#tuning-the-hyper-parameters-of-the-quantile-regressors">Tuning the hyper-parameters of the quantile regressors</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div>
<div id="sk-page-content-wrapper">
<div class="sk-page-content container-fluid body px-md-3" role="main">
<div class="sphx-glr-download-link-note admonition note">
<p class="admonition-title">Note</p>
<p><a class="reference internal" href="#sphx-glr-download-auto-examples-ensemble-plot-gradient-boosting-quantile-py"><span class="std std-ref">Go to the end</span></a>
to download the full example code or to run this example in your browser via JupyterLite or Binder</p>
</div>
<section class="sphx-glr-example-title" id="prediction-intervals-for-gradient-boosting-regression">
<span id="sphx-glr-auto-examples-ensemble-plot-gradient-boosting-quantile-py"></span><h1>Prediction Intervals for Gradient Boosting Regression<a class="headerlink" href="#prediction-intervals-for-gradient-boosting-regression" title="Permalink to this heading">¶</a></h1>
<p>This example shows how quantile regression can be used to create prediction
intervals.</p>
<p>Generate some data for a synthetic regression problem by applying the
function f to uniformly sampled random inputs.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="kn">from</span> <span class="nn">sklearn.model_selection</span> <span class="kn">import</span> <a href="../../modules/generated/sklearn.model_selection.train_test_split.html#sklearn.model_selection.train_test_split" title="sklearn.model_selection.train_test_split" class="sphx-glr-backref-module-sklearn-model_selection sphx-glr-backref-type-py-function"><span class="n">train_test_split</span></a>
<span class="k">def</span> <span class="nf">f</span><span class="p">(</span><span class="n">x</span><span class="p">):</span>
<span class="w"> </span><span class="sd">"""The function to predict."""</span>
<span class="k">return</span> <span class="n">x</span> <span class="o">*</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.sin.html#numpy.sin" title="numpy.sin" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-data"><span class="n">np</span><span class="o">.</span><span class="n">sin</span></a><span class="p">(</span><span class="n">x</span><span class="p">)</span>
<span class="n">rng</span> <span class="o">=</span> <a href="https://numpy.org/doc/stable/reference/random/legacy.html#numpy.random.RandomState" title="numpy.random.RandomState" class="sphx-glr-backref-module-numpy-random sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">RandomState</span></a><span class="p">(</span><span class="mi">42</span><span class="p">)</span>
<span class="n">X</span> <span class="o">=</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.atleast_2d.html#numpy.atleast_2d" title="numpy.atleast_2d" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-function"><span class="n">np</span><span class="o">.</span><span class="n">atleast_2d</span></a><span class="p">(</span><span class="n">rng</span><span class="o">.</span><span class="n">uniform</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mf">10.0</span><span class="p">,</span> <span class="n">size</span><span class="o">=</span><span class="mi">1000</span><span class="p">))</span><span class="o">.</span><span class="n">T</span>
<span class="n">expected_y</span> <span class="o">=</span> <span class="n">f</span><span class="p">(</span><span class="n">X</span><span class="p">)</span><span class="o">.</span><span class="n">ravel</span><span class="p">()</span>
</pre></div>
</div>
<p>To make the problem interesting, we generate observations of the target y as
the sum of a deterministic term computed by the function f and a random noise
term that follows a centered <a class="reference external" href="https://en.wikipedia.org/wiki/Log-normal_distribution">log-normal</a>. To make this even
more interesting we consider the case where the amplitude of the noise
depends on the input variable x (heteroscedastic noise).</p>
<p>The lognormal distribution is non-symmetric and long tailed: observing large
outliers is likely but it is impossible to observe small outliers.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">sigma</span> <span class="o">=</span> <span class="mf">0.5</span> <span class="o">+</span> <span class="n">X</span><span class="o">.</span><span class="n">ravel</span><span class="p">()</span> <span class="o">/</span> <span class="mi">10</span>
<span class="n">noise</span> <span class="o">=</span> <span class="n">rng</span><span class="o">.</span><span class="n">lognormal</span><span class="p">(</span><span class="n">sigma</span><span class="o">=</span><span class="n">sigma</span><span class="p">)</span> <span class="o">-</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.exp.html#numpy.exp" title="numpy.exp" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-data"><span class="n">np</span><span class="o">.</span><span class="n">exp</span></a><span class="p">(</span><span class="n">sigma</span><span class="o">**</span><span class="mi">2</span> <span class="o">/</span> <span class="mi">2</span><span class="p">)</span>
<span class="n">y</span> <span class="o">=</span> <span class="n">expected_y</span> <span class="o">+</span> <span class="n">noise</span>
</pre></div>
</div>
<p>Split into train, test datasets:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">X_train</span><span class="p">,</span> <span class="n">X_test</span><span class="p">,</span> <span class="n">y_train</span><span class="p">,</span> <span class="n">y_test</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.model_selection.train_test_split.html#sklearn.model_selection.train_test_split" title="sklearn.model_selection.train_test_split" class="sphx-glr-backref-module-sklearn-model_selection sphx-glr-backref-type-py-function"><span class="n">train_test_split</span></a><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
</pre></div>
</div>
<section id="fitting-non-linear-quantile-and-least-squares-regressors">
<h2>Fitting non-linear quantile and least squares regressors<a class="headerlink" href="#fitting-non-linear-quantile-and-least-squares-regressors" title="Permalink to this heading">¶</a></h2>
<p>Fit gradient boosting models trained with the quantile loss and
alpha=0.05, 0.5, 0.95.</p>
<p>The models obtained for alpha=0.05 and alpha=0.95 produce a 90% confidence
interval (95% - 5% = 90%).</p>
<p>The model trained with alpha=0.5 produces a regression of the median: on
average, there should be the same number of target observations above and
below the predicted values.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">sklearn.ensemble</span> <span class="kn">import</span> <a href="../../modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor" class="sphx-glr-backref-module-sklearn-ensemble sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">GradientBoostingRegressor</span></a>
<span class="kn">from</span> <span class="nn">sklearn.metrics</span> <span class="kn">import</span> <a href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_pinball_loss</span></a><span class="p">,</span> <a href="../../modules/generated/sklearn.metrics.mean_squared_error.html#sklearn.metrics.mean_squared_error" title="sklearn.metrics.mean_squared_error" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_squared_error</span></a>
<span class="n">all_models</span> <span class="o">=</span> <span class="p">{}</span>
<span class="n">common_params</span> <span class="o">=</span> <span class="nb">dict</span><span class="p">(</span>
<span class="n">learning_rate</span><span class="o">=</span><span class="mf">0.05</span><span class="p">,</span>
<span class="n">n_estimators</span><span class="o">=</span><span class="mi">200</span><span class="p">,</span>
<span class="n">max_depth</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span>
<span class="n">min_samples_leaf</span><span class="o">=</span><span class="mi">9</span><span class="p">,</span>
<span class="n">min_samples_split</span><span class="o">=</span><span class="mi">9</span><span class="p">,</span>
<span class="p">)</span>
<span class="k">for</span> <span class="n">alpha</span> <span class="ow">in</span> <span class="p">[</span><span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.95</span><span class="p">]:</span>
<span class="n">gbr</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor" class="sphx-glr-backref-module-sklearn-ensemble sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">GradientBoostingRegressor</span></a><span class="p">(</span><span class="n">loss</span><span class="o">=</span><span class="s2">"quantile"</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">,</span> <span class="o">**</span><span class="n">common_params</span><span class="p">)</span>
<span class="n">all_models</span><span class="p">[</span><span class="s2">"q </span><span class="si">%1.2f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">alpha</span><span class="p">]</span> <span class="o">=</span> <span class="n">gbr</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
</pre></div>
</div>
<p>Notice that <a class="reference internal" href="../../modules/generated/sklearn.ensemble.HistGradientBoostingRegressor.html#sklearn.ensemble.HistGradientBoostingRegressor" title="sklearn.ensemble.HistGradientBoostingRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">HistGradientBoostingRegressor</span></code></a> is much
faster than <a class="reference internal" href="../../modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor"><code class="xref py py-class docutils literal notranslate"><span class="pre">GradientBoostingRegressor</span></code></a> starting with
intermediate datasets (<code class="docutils literal notranslate"><span class="pre">n_samples</span> <span class="pre">>=</span> <span class="pre">10_000</span></code>), which is not the case of the
present example.</p>
<p>For the sake of comparison, we also fit a baseline model trained with the
usual (mean) squared error (MSE).</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">gbr_ls</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor" class="sphx-glr-backref-module-sklearn-ensemble sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">GradientBoostingRegressor</span></a><span class="p">(</span><span class="n">loss</span><span class="o">=</span><span class="s2">"squared_error"</span><span class="p">,</span> <span class="o">**</span><span class="n">common_params</span><span class="p">)</span>
<span class="n">all_models</span><span class="p">[</span><span class="s2">"mse"</span><span class="p">]</span> <span class="o">=</span> <span class="n">gbr_ls</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
</pre></div>
</div>
<p>Create an evenly spaced evaluation set of input values spanning the [0, 10]
range.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">xx</span> <span class="o">=</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.atleast_2d.html#numpy.atleast_2d" title="numpy.atleast_2d" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-function"><span class="n">np</span><span class="o">.</span><span class="n">atleast_2d</span></a><span class="p">(</span><a href="https://numpy.org/doc/stable/reference/generated/numpy.linspace.html#numpy.linspace" title="numpy.linspace" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-function"><span class="n">np</span><span class="o">.</span><span class="n">linspace</span></a><span class="p">(</span><span class="mi">0</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">1000</span><span class="p">))</span><span class="o">.</span><span class="n">T</span>
</pre></div>
</div>
<p>Plot the true conditional mean function f, the predictions of the conditional
mean (loss equals squared error), the conditional median and the conditional
90% interval (from 5th to 95th conditional percentiles).</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">matplotlib.pyplot</span> <span class="k">as</span> <span class="nn">plt</span>
<span class="n">y_pred</span> <span class="o">=</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"mse"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">y_lower</span> <span class="o">=</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.05"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">y_upper</span> <span class="o">=</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.95"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">y_med</span> <span class="o">=</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.50"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">fig</span> <span class="o">=</span> <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.figure.html#matplotlib.pyplot.figure" title="matplotlib.pyplot.figure" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">figure</span></a><span class="p">(</span><span class="n">figsize</span><span class="o">=</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="mi">10</span><span class="p">))</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">f</span><span class="p">(</span><span class="n">xx</span><span class="p">),</span> <span class="s2">"g:"</span><span class="p">,</span> <span class="n">linewidth</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="sa">r</span><span class="s2">"$f(x) = x\,\sin(x)$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">,</span> <span class="s2">"b."</span><span class="p">,</span> <span class="n">markersize</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Test observations"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_med</span><span class="p">,</span> <span class="s2">"r-"</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Predicted median"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_pred</span><span class="p">,</span> <span class="s2">"r-"</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Predicted mean"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_upper</span><span class="p">,</span> <span class="s2">"k-"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_lower</span><span class="p">,</span> <span class="s2">"k-"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.fill_between.html#matplotlib.pyplot.fill_between" title="matplotlib.pyplot.fill_between" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">fill_between</span></a><span class="p">(</span>
<span class="n">xx</span><span class="o">.</span><span class="n">ravel</span><span class="p">(),</span> <span class="n">y_lower</span><span class="p">,</span> <span class="n">y_upper</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.4</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Predicted 90</span><span class="si">% i</span><span class="s2">nterval"</span>
<span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.xlabel.html#matplotlib.pyplot.xlabel" title="matplotlib.pyplot.xlabel" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">xlabel</span></a><span class="p">(</span><span class="s2">"$x$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.ylabel.html#matplotlib.pyplot.ylabel" title="matplotlib.pyplot.ylabel" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">ylabel</span></a><span class="p">(</span><span class="s2">"$f(x)$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.ylim.html#matplotlib.pyplot.ylim" title="matplotlib.pyplot.ylim" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">ylim</span></a><span class="p">(</span><span class="o">-</span><span class="mi">10</span><span class="p">,</span> <span class="mi">25</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.legend.html#matplotlib.pyplot.legend" title="matplotlib.pyplot.legend" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">legend</span></a><span class="p">(</span><span class="n">loc</span><span class="o">=</span><span class="s2">"upper left"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.show.html#matplotlib.pyplot.show" title="matplotlib.pyplot.show" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">show</span></a><span class="p">()</span>
</pre></div>
</div>
<img src="../../_images/sphx_glr_plot_gradient_boosting_quantile_001.png" srcset="../../_images/sphx_glr_plot_gradient_boosting_quantile_001.png" alt="plot gradient boosting quantile" class = "sphx-glr-single-img"/><p>Comparing the predicted median with the predicted mean, we note that the
median is on average below the mean as the noise is skewed towards high
values (large outliers). The median estimate also seems to be smoother
because of its natural robustness to outliers.</p>
<p>Also observe that the inductive bias of gradient boosting trees is
unfortunately preventing our 0.05 quantile to fully capture the sinoisoidal
shape of the signal, in particular around x=8. Tuning hyper-parameters can
reduce this effect as shown in the last part of this notebook.</p>
</section>
<section id="analysis-of-the-error-metrics">
<h2>Analysis of the error metrics<a class="headerlink" href="#analysis-of-the-error-metrics" title="Permalink to this heading">¶</a></h2>
<p>Measure the models with <a class="reference internal" href="../../modules/generated/sklearn.metrics.mean_squared_error.html#sklearn.metrics.mean_squared_error" title="sklearn.metrics.mean_squared_error"><code class="xref py py-func docutils literal notranslate"><span class="pre">mean_squared_error</span></code></a> and
<a class="reference internal" href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss"><code class="xref py py-func docutils literal notranslate"><span class="pre">mean_pinball_loss</span></code></a> metrics on the training dataset.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">pandas</span> <span class="k">as</span> <span class="nn">pd</span>
<span class="k">def</span> <span class="nf">highlight_min</span><span class="p">(</span><span class="n">x</span><span class="p">):</span>
<span class="n">x_min</span> <span class="o">=</span> <span class="n">x</span><span class="o">.</span><span class="n">min</span><span class="p">()</span>
<span class="k">return</span> <span class="p">[</span><span class="s2">"font-weight: bold"</span> <span class="k">if</span> <span class="n">v</span> <span class="o">==</span> <span class="n">x_min</span> <span class="k">else</span> <span class="s2">""</span> <span class="k">for</span> <span class="n">v</span> <span class="ow">in</span> <span class="n">x</span><span class="p">]</span>
<span class="n">results</span> <span class="o">=</span> <span class="p">[]</span>
<span class="k">for</span> <span class="n">name</span><span class="p">,</span> <span class="n">gbr</span> <span class="ow">in</span> <span class="nb">sorted</span><span class="p">(</span><span class="n">all_models</span><span class="o">.</span><span class="n">items</span><span class="p">()):</span>
<span class="n">metrics</span> <span class="o">=</span> <span class="p">{</span><span class="s2">"model"</span><span class="p">:</span> <span class="n">name</span><span class="p">}</span>
<span class="n">y_pred</span> <span class="o">=</span> <span class="n">gbr</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_train</span><span class="p">)</span>
<span class="k">for</span> <span class="n">alpha</span> <span class="ow">in</span> <span class="p">[</span><span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.95</span><span class="p">]:</span>
<span class="n">metrics</span><span class="p">[</span><span class="s2">"pbl=</span><span class="si">%1.2f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">alpha</span><span class="p">]</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_pinball_loss</span></a><span class="p">(</span><span class="n">y_train</span><span class="p">,</span> <span class="n">y_pred</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">)</span>
<span class="n">metrics</span><span class="p">[</span><span class="s2">"MSE"</span><span class="p">]</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.mean_squared_error.html#sklearn.metrics.mean_squared_error" title="sklearn.metrics.mean_squared_error" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_squared_error</span></a><span class="p">(</span><span class="n">y_train</span><span class="p">,</span> <span class="n">y_pred</span><span class="p">)</span>
<span class="n">results</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">metrics</span><span class="p">)</span>
<a href="https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.html#pandas.DataFrame" title="pandas.DataFrame" class="sphx-glr-backref-module-pandas sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">pd</span><span class="o">.</span><span class="n">DataFrame</span></a><span class="p">(</span><span class="n">results</span><span class="p">)</span><span class="o">.</span><span class="n">set_index</span><span class="p">(</span><span class="s2">"model"</span><span class="p">)</span><span class="o">.</span><span class="n">style</span><span class="o">.</span><span class="n">apply</span><span class="p">(</span><span class="n">highlight_min</span><span class="p">)</span>
</pre></div>
</div>
<div class="output_subarea output_html rendered_html output_result">
<style type="text/css">
#T_59424_row0_col3, #T_59424_row1_col0, #T_59424_row2_col1, #T_59424_row3_col2 {
font-weight: bold;
}
</style>
<table id="T_59424">
<thead>
<tr>
<th class="blank level0" > </th>
<th id="T_59424_level0_col0" class="col_heading level0 col0" >pbl=0.05</th>
<th id="T_59424_level0_col1" class="col_heading level0 col1" >pbl=0.50</th>
<th id="T_59424_level0_col2" class="col_heading level0 col2" >pbl=0.95</th>
<th id="T_59424_level0_col3" class="col_heading level0 col3" >MSE</th>
</tr>
<tr>
<th class="index_name level0" >model</th>
<th class="blank col0" > </th>
<th class="blank col1" > </th>
<th class="blank col2" > </th>
<th class="blank col3" > </th>
</tr>
</thead>
<tbody>
<tr>
<th id="T_59424_level0_row0" class="row_heading level0 row0" >mse</th>
<td id="T_59424_row0_col0" class="data row0 col0" >0.715413</td>
<td id="T_59424_row0_col1" class="data row0 col1" >0.715413</td>
<td id="T_59424_row0_col2" class="data row0 col2" >0.715413</td>
<td id="T_59424_row0_col3" class="data row0 col3" >7.750348</td>
</tr>
<tr>
<th id="T_59424_level0_row1" class="row_heading level0 row1" >q 0.05</th>
<td id="T_59424_row1_col0" class="data row1 col0" >0.127128</td>
<td id="T_59424_row1_col1" class="data row1 col1" >1.253445</td>
<td id="T_59424_row1_col2" class="data row1 col2" >2.379763</td>
<td id="T_59424_row1_col3" class="data row1 col3" >18.933253</td>
</tr>
<tr>
<th id="T_59424_level0_row2" class="row_heading level0 row2" >q 0.50</th>
<td id="T_59424_row2_col0" class="data row2 col0" >0.305438</td>
<td id="T_59424_row2_col1" class="data row2 col1" >0.622811</td>
<td id="T_59424_row2_col2" class="data row2 col2" >0.940184</td>
<td id="T_59424_row2_col3" class="data row2 col3" >9.827917</td>
</tr>
<tr>
<th id="T_59424_level0_row3" class="row_heading level0 row3" >q 0.95</th>
<td id="T_59424_row3_col0" class="data row3 col0" >3.909909</td>
<td id="T_59424_row3_col1" class="data row3 col1" >2.145957</td>
<td id="T_59424_row3_col2" class="data row3 col2" >0.382005</td>
<td id="T_59424_row3_col3" class="data row3 col3" >28.667219</td>
</tr>
</tbody>
</table>
</div>
<br />
<br /><p>One column shows all models evaluated by the same metric. The minimum number
on a column should be obtained when the model is trained and measured with
the same metric. This should be always the case on the training set if the
training converged.</p>
<p>Note that because the target distribution is asymmetric, the expected
conditional mean and conditional median are significantly different and
therefore one could not use the squared error model get a good estimation of
the conditional median nor the converse.</p>
<p>If the target distribution were symmetric and had no outliers (e.g. with a
Gaussian noise), then median estimator and the least squares estimator would
have yielded similar predictions.</p>
<p>We then do the same on the test set.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">results</span> <span class="o">=</span> <span class="p">[]</span>
<span class="k">for</span> <span class="n">name</span><span class="p">,</span> <span class="n">gbr</span> <span class="ow">in</span> <span class="nb">sorted</span><span class="p">(</span><span class="n">all_models</span><span class="o">.</span><span class="n">items</span><span class="p">()):</span>
<span class="n">metrics</span> <span class="o">=</span> <span class="p">{</span><span class="s2">"model"</span><span class="p">:</span> <span class="n">name</span><span class="p">}</span>
<span class="n">y_pred</span> <span class="o">=</span> <span class="n">gbr</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span>
<span class="k">for</span> <span class="n">alpha</span> <span class="ow">in</span> <span class="p">[</span><span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.95</span><span class="p">]:</span>
<span class="n">metrics</span><span class="p">[</span><span class="s2">"pbl=</span><span class="si">%1.2f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">alpha</span><span class="p">]</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_pinball_loss</span></a><span class="p">(</span><span class="n">y_test</span><span class="p">,</span> <span class="n">y_pred</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">)</span>
<span class="n">metrics</span><span class="p">[</span><span class="s2">"MSE"</span><span class="p">]</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.mean_squared_error.html#sklearn.metrics.mean_squared_error" title="sklearn.metrics.mean_squared_error" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_squared_error</span></a><span class="p">(</span><span class="n">y_test</span><span class="p">,</span> <span class="n">y_pred</span><span class="p">)</span>
<span class="n">results</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">metrics</span><span class="p">)</span>
<a href="https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.html#pandas.DataFrame" title="pandas.DataFrame" class="sphx-glr-backref-module-pandas sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">pd</span><span class="o">.</span><span class="n">DataFrame</span></a><span class="p">(</span><span class="n">results</span><span class="p">)</span><span class="o">.</span><span class="n">set_index</span><span class="p">(</span><span class="s2">"model"</span><span class="p">)</span><span class="o">.</span><span class="n">style</span><span class="o">.</span><span class="n">apply</span><span class="p">(</span><span class="n">highlight_min</span><span class="p">)</span>
</pre></div>
</div>
<div class="output_subarea output_html rendered_html output_result">
<style type="text/css">
#T_14b8e_row1_col0, #T_14b8e_row2_col1, #T_14b8e_row2_col3, #T_14b8e_row3_col2 {
font-weight: bold;
}
</style>
<table id="T_14b8e">
<thead>
<tr>
<th class="blank level0" > </th>
<th id="T_14b8e_level0_col0" class="col_heading level0 col0" >pbl=0.05</th>
<th id="T_14b8e_level0_col1" class="col_heading level0 col1" >pbl=0.50</th>
<th id="T_14b8e_level0_col2" class="col_heading level0 col2" >pbl=0.95</th>
<th id="T_14b8e_level0_col3" class="col_heading level0 col3" >MSE</th>
</tr>
<tr>
<th class="index_name level0" >model</th>
<th class="blank col0" > </th>
<th class="blank col1" > </th>
<th class="blank col2" > </th>
<th class="blank col3" > </th>
</tr>
</thead>
<tbody>
<tr>
<th id="T_14b8e_level0_row0" class="row_heading level0 row0" >mse</th>
<td id="T_14b8e_row0_col0" class="data row0 col0" >0.917281</td>
<td id="T_14b8e_row0_col1" class="data row0 col1" >0.767498</td>
<td id="T_14b8e_row0_col2" class="data row0 col2" >0.617715</td>
<td id="T_14b8e_row0_col3" class="data row0 col3" >6.692901</td>
</tr>
<tr>
<th id="T_14b8e_level0_row1" class="row_heading level0 row1" >q 0.05</th>
<td id="T_14b8e_row1_col0" class="data row1 col0" >0.144204</td>
<td id="T_14b8e_row1_col1" class="data row1 col1" >1.245961</td>
<td id="T_14b8e_row1_col2" class="data row1 col2" >2.347717</td>
<td id="T_14b8e_row1_col3" class="data row1 col3" >15.648026</td>
</tr>
<tr>
<th id="T_14b8e_level0_row2" class="row_heading level0 row2" >q 0.50</th>
<td id="T_14b8e_row2_col0" class="data row2 col0" >0.412021</td>
<td id="T_14b8e_row2_col1" class="data row2 col1" >0.607752</td>
<td id="T_14b8e_row2_col2" class="data row2 col2" >0.803483</td>
<td id="T_14b8e_row2_col3" class="data row2 col3" >5.874771</td>
</tr>
<tr>
<th id="T_14b8e_level0_row3" class="row_heading level0 row3" >q 0.95</th>
<td id="T_14b8e_row3_col0" class="data row3 col0" >4.354394</td>
<td id="T_14b8e_row3_col1" class="data row3 col1" >2.355445</td>
<td id="T_14b8e_row3_col2" class="data row3 col2" >0.356497</td>
<td id="T_14b8e_row3_col3" class="data row3 col3" >34.852774</td>
</tr>
</tbody>
</table>
</div>
<br />
<br /><p>Errors are higher meaning the models slightly overfitted the data. It still
shows that the best test metric is obtained when the model is trained by
minimizing this same metric.</p>
<p>Note that the conditional median estimator is competitive with the squared
error estimator in terms of MSE on the test set: this can be explained by
the fact the squared error estimator is very sensitive to large outliers
which can cause significant overfitting. This can be seen on the right hand
side of the previous plot. The conditional median estimator is biased
(underestimation for this asymmetric noise) but is also naturally robust to
outliers and overfits less.</p>
</section>
<section id="calibration-of-the-confidence-interval">
<h2>Calibration of the confidence interval<a class="headerlink" href="#calibration-of-the-confidence-interval" title="Permalink to this heading">¶</a></h2>
<p>We can also evaluate the ability of the two extreme quantile estimators at
producing a well-calibrated conditional 90%-confidence interval.</p>
<p>To do this we can compute the fraction of observations that fall between the
predictions:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">def</span> <span class="nf">coverage_fraction</span><span class="p">(</span><span class="n">y</span><span class="p">,</span> <span class="n">y_low</span><span class="p">,</span> <span class="n">y_high</span><span class="p">):</span>
<span class="k">return</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.mean.html#numpy.mean" title="numpy.mean" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-function"><span class="n">np</span><span class="o">.</span><span class="n">mean</span></a><span class="p">(</span><a href="https://numpy.org/doc/stable/reference/generated/numpy.logical_and.html#numpy.logical_and" title="numpy.logical_and" class="sphx-glr-backref-module-numpy sphx-glr-backref-type-py-data"><span class="n">np</span><span class="o">.</span><span class="n">logical_and</span></a><span class="p">(</span><span class="n">y</span> <span class="o">>=</span> <span class="n">y_low</span><span class="p">,</span> <span class="n">y</span> <span class="o"><=</span> <span class="n">y_high</span><span class="p">))</span>
<span class="n">coverage_fraction</span><span class="p">(</span>
<span class="n">y_train</span><span class="p">,</span>
<span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.05"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_train</span><span class="p">),</span>
<span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.95"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_train</span><span class="p">),</span>
<span class="p">)</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>0.9
</pre></div>
</div>
<p>On the training set the calibration is very close to the expected coverage
value for a 90% confidence interval.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">coverage_fraction</span><span class="p">(</span>
<span class="n">y_test</span><span class="p">,</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.05"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">),</span> <span class="n">all_models</span><span class="p">[</span><span class="s2">"q 0.95"</span><span class="p">]</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">)</span>
<span class="p">)</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>0.868
</pre></div>
</div>
<p>On the test set, the estimated confidence interval is slightly too narrow.
Note, however, that we would need to wrap those metrics in a cross-validation
loop to assess their variability under data resampling.</p>
</section>
<section id="tuning-the-hyper-parameters-of-the-quantile-regressors">
<h2>Tuning the hyper-parameters of the quantile regressors<a class="headerlink" href="#tuning-the-hyper-parameters-of-the-quantile-regressors" title="Permalink to this heading">¶</a></h2>
<p>In the plot above, we observed that the 5th percentile regressor seems to
underfit and could not adapt to sinusoidal shape of the signal.</p>
<p>The hyper-parameters of the model were approximately hand-tuned for the
median regressor and there is no reason that the same hyper-parameters are
suitable for the 5th percentile regressor.</p>
<p>To confirm this hypothesis, we tune the hyper-parameters of a new regressor
of the 5th percentile by selecting the best model parameters by
cross-validation on the pinball loss with alpha=0.05:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">sklearn.experimental</span> <span class="kn">import</span> <span class="n">enable_halving_search_cv</span> <span class="c1"># noqa</span>
<span class="kn">from</span> <span class="nn">sklearn.model_selection</span> <span class="kn">import</span> <a href="../../modules/generated/sklearn.model_selection.HalvingRandomSearchCV.html#sklearn.model_selection.HalvingRandomSearchCV" title="sklearn.model_selection.HalvingRandomSearchCV" class="sphx-glr-backref-module-sklearn-model_selection sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">HalvingRandomSearchCV</span></a>
<span class="kn">from</span> <span class="nn">sklearn.metrics</span> <span class="kn">import</span> <a href="../../modules/generated/sklearn.metrics.make_scorer.html#sklearn.metrics.make_scorer" title="sklearn.metrics.make_scorer" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">make_scorer</span></a>
<span class="kn">from</span> <span class="nn">pprint</span> <span class="kn">import</span> <a href="https://docs.python.org/3/library/pprint.html#pprint.pprint" title="pprint.pprint" class="sphx-glr-backref-module-pprint sphx-glr-backref-type-py-function"><span class="n">pprint</span></a>
<span class="n">param_grid</span> <span class="o">=</span> <span class="nb">dict</span><span class="p">(</span>
<span class="n">learning_rate</span><span class="o">=</span><span class="p">[</span><span class="mf">0.05</span><span class="p">,</span> <span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">],</span>
<span class="n">max_depth</span><span class="o">=</span><span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">],</span>
<span class="n">min_samples_leaf</span><span class="o">=</span><span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">20</span><span class="p">],</span>
<span class="n">min_samples_split</span><span class="o">=</span><span class="p">[</span><span class="mi">5</span><span class="p">,</span> <span class="mi">10</span><span class="p">,</span> <span class="mi">20</span><span class="p">,</span> <span class="mi">30</span><span class="p">,</span> <span class="mi">50</span><span class="p">],</span>
<span class="p">)</span>
<span class="n">alpha</span> <span class="o">=</span> <span class="mf">0.05</span>
<span class="n">neg_mean_pinball_loss_05p_scorer</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.make_scorer.html#sklearn.metrics.make_scorer" title="sklearn.metrics.make_scorer" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">make_scorer</span></a><span class="p">(</span>
<a href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_pinball_loss</span></a><span class="p">,</span>
<span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">,</span>
<span class="n">greater_is_better</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="c1"># maximize the negative loss</span>
<span class="p">)</span>
<span class="n">gbr</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.ensemble.GradientBoostingRegressor.html#sklearn.ensemble.GradientBoostingRegressor" title="sklearn.ensemble.GradientBoostingRegressor" class="sphx-glr-backref-module-sklearn-ensemble sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">GradientBoostingRegressor</span></a><span class="p">(</span><span class="n">loss</span><span class="o">=</span><span class="s2">"quantile"</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">,</span> <span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="n">search_05p</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.model_selection.HalvingRandomSearchCV.html#sklearn.model_selection.HalvingRandomSearchCV" title="sklearn.model_selection.HalvingRandomSearchCV" class="sphx-glr-backref-module-sklearn-model_selection sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">HalvingRandomSearchCV</span></a><span class="p">(</span>
<span class="n">gbr</span><span class="p">,</span>
<span class="n">param_grid</span><span class="p">,</span>
<span class="n">resource</span><span class="o">=</span><span class="s2">"n_estimators"</span><span class="p">,</span>
<span class="n">max_resources</span><span class="o">=</span><span class="mi">250</span><span class="p">,</span>
<span class="n">min_resources</span><span class="o">=</span><span class="mi">50</span><span class="p">,</span>
<span class="n">scoring</span><span class="o">=</span><span class="n">neg_mean_pinball_loss_05p_scorer</span><span class="p">,</span>
<span class="n">n_jobs</span><span class="o">=</span><span class="mi">2</span><span class="p">,</span>
<span class="n">random_state</span><span class="o">=</span><span class="mi">0</span><span class="p">,</span>
<span class="p">)</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<a href="https://docs.python.org/3/library/pprint.html#pprint.pprint" title="pprint.pprint" class="sphx-glr-backref-module-pprint sphx-glr-backref-type-py-function"><span class="n">pprint</span></a><span class="p">(</span><span class="n">search_05p</span><span class="o">.</span><span class="n">best_params_</span><span class="p">)</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>{'learning_rate': 0.2,
'max_depth': 2,
'min_samples_leaf': 20,
'min_samples_split': 10,
'n_estimators': 150}
</pre></div>
</div>
<p>We observe that the hyper-parameters that were hand-tuned for the median
regressor are in the same range as the hyper-parameters suitable for the 5th
percentile regressor.</p>
<p>Let’s now tune the hyper-parameters for the 95th percentile regressor. We
need to redefine the <code class="docutils literal notranslate"><span class="pre">scoring</span></code> metric used to select the best model, along
with adjusting the alpha parameter of the inner gradient boosting estimator
itself:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">sklearn.base</span> <span class="kn">import</span> <span class="n">clone</span>
<span class="n">alpha</span> <span class="o">=</span> <span class="mf">0.95</span>
<span class="n">neg_mean_pinball_loss_95p_scorer</span> <span class="o">=</span> <a href="../../modules/generated/sklearn.metrics.make_scorer.html#sklearn.metrics.make_scorer" title="sklearn.metrics.make_scorer" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">make_scorer</span></a><span class="p">(</span>
<a href="../../modules/generated/sklearn.metrics.mean_pinball_loss.html#sklearn.metrics.mean_pinball_loss" title="sklearn.metrics.mean_pinball_loss" class="sphx-glr-backref-module-sklearn-metrics sphx-glr-backref-type-py-function"><span class="n">mean_pinball_loss</span></a><span class="p">,</span>
<span class="n">alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">,</span>
<span class="n">greater_is_better</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span> <span class="c1"># maximize the negative loss</span>
<span class="p">)</span>
<span class="n">search_95p</span> <span class="o">=</span> <span class="n">clone</span><span class="p">(</span><span class="n">search_05p</span><span class="p">)</span><span class="o">.</span><span class="n">set_params</span><span class="p">(</span>
<span class="n">estimator__alpha</span><span class="o">=</span><span class="n">alpha</span><span class="p">,</span>
<span class="n">scoring</span><span class="o">=</span><span class="n">neg_mean_pinball_loss_95p_scorer</span><span class="p">,</span>
<span class="p">)</span>
<span class="n">search_95p</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X_train</span><span class="p">,</span> <span class="n">y_train</span><span class="p">)</span>
<a href="https://docs.python.org/3/library/pprint.html#pprint.pprint" title="pprint.pprint" class="sphx-glr-backref-module-pprint sphx-glr-backref-type-py-function"><span class="n">pprint</span></a><span class="p">(</span><span class="n">search_95p</span><span class="o">.</span><span class="n">best_params_</span><span class="p">)</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>{'learning_rate': 0.05,
'max_depth': 2,
'min_samples_leaf': 5,
'min_samples_split': 20,
'n_estimators': 150}
</pre></div>
</div>
<p>The result shows that the hyper-parameters for the 95th percentile regressor
identified by the search procedure are roughly in the same range as the hand-
tuned hyper-parameters for the median regressor and the hyper-parameters
identified by the search procedure for the 5th percentile regressor. However,
the hyper-parameter searches did lead to an improved 90% confidence interval
that is comprised by the predictions of those two tuned quantile regressors.
Note that the prediction of the upper 95th percentile has a much coarser shape
than the prediction of the lower 5th percentile because of the outliers:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">y_lower</span> <span class="o">=</span> <span class="n">search_05p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">y_upper</span> <span class="o">=</span> <span class="n">search_95p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">xx</span><span class="p">)</span>
<span class="n">fig</span> <span class="o">=</span> <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.figure.html#matplotlib.pyplot.figure" title="matplotlib.pyplot.figure" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">figure</span></a><span class="p">(</span><span class="n">figsize</span><span class="o">=</span><span class="p">(</span><span class="mi">10</span><span class="p">,</span> <span class="mi">10</span><span class="p">))</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">f</span><span class="p">(</span><span class="n">xx</span><span class="p">),</span> <span class="s2">"g:"</span><span class="p">,</span> <span class="n">linewidth</span><span class="o">=</span><span class="mi">3</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="sa">r</span><span class="s2">"$f(x) = x\,\sin(x)$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">X_test</span><span class="p">,</span> <span class="n">y_test</span><span class="p">,</span> <span class="s2">"b."</span><span class="p">,</span> <span class="n">markersize</span><span class="o">=</span><span class="mi">10</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Test observations"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_upper</span><span class="p">,</span> <span class="s2">"k-"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.plot.html#matplotlib.pyplot.plot" title="matplotlib.pyplot.plot" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">plot</span></a><span class="p">(</span><span class="n">xx</span><span class="p">,</span> <span class="n">y_lower</span><span class="p">,</span> <span class="s2">"k-"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.fill_between.html#matplotlib.pyplot.fill_between" title="matplotlib.pyplot.fill_between" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">fill_between</span></a><span class="p">(</span>
<span class="n">xx</span><span class="o">.</span><span class="n">ravel</span><span class="p">(),</span> <span class="n">y_lower</span><span class="p">,</span> <span class="n">y_upper</span><span class="p">,</span> <span class="n">alpha</span><span class="o">=</span><span class="mf">0.4</span><span class="p">,</span> <span class="n">label</span><span class="o">=</span><span class="s2">"Predicted 90</span><span class="si">% i</span><span class="s2">nterval"</span>
<span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.xlabel.html#matplotlib.pyplot.xlabel" title="matplotlib.pyplot.xlabel" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">xlabel</span></a><span class="p">(</span><span class="s2">"$x$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.ylabel.html#matplotlib.pyplot.ylabel" title="matplotlib.pyplot.ylabel" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">ylabel</span></a><span class="p">(</span><span class="s2">"$f(x)$"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.ylim.html#matplotlib.pyplot.ylim" title="matplotlib.pyplot.ylim" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">ylim</span></a><span class="p">(</span><span class="o">-</span><span class="mi">10</span><span class="p">,</span> <span class="mi">25</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.legend.html#matplotlib.pyplot.legend" title="matplotlib.pyplot.legend" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">legend</span></a><span class="p">(</span><span class="n">loc</span><span class="o">=</span><span class="s2">"upper left"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.title.html#matplotlib.pyplot.title" title="matplotlib.pyplot.title" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">title</span></a><span class="p">(</span><span class="s2">"Prediction with tuned hyper-parameters"</span><span class="p">)</span>
<a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.show.html#matplotlib.pyplot.show" title="matplotlib.pyplot.show" class="sphx-glr-backref-module-matplotlib-pyplot sphx-glr-backref-type-py-function"><span class="n">plt</span><span class="o">.</span><span class="n">show</span></a><span class="p">()</span>
</pre></div>
</div>
<img src="../../_images/sphx_glr_plot_gradient_boosting_quantile_002.png" srcset="../../_images/sphx_glr_plot_gradient_boosting_quantile_002.png" alt="Prediction with tuned hyper-parameters" class = "sphx-glr-single-img"/><p>The plot looks qualitatively better than for the untuned models, especially
for the shape of the of lower quantile.</p>
<p>We now quantitatively evaluate the joint-calibration of the pair of
estimators:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">coverage_fraction</span><span class="p">(</span><span class="n">y_train</span><span class="p">,</span> <span class="n">search_05p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_train</span><span class="p">),</span> <span class="n">search_95p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_train</span><span class="p">))</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>0.9026666666666666
</pre></div>
</div>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">coverage_fraction</span><span class="p">(</span><span class="n">y_test</span><span class="p">,</span> <span class="n">search_05p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">),</span> <span class="n">search_95p</span><span class="o">.</span><span class="n">predict</span><span class="p">(</span><span class="n">X_test</span><span class="p">))</span>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>0.796
</pre></div>
</div>
<p>The calibration of the tuned pair is sadly not better on the test set: the
width of the estimated confidence interval is still too narrow.</p>
<p>Again, we would need to wrap this study in a cross-validation loop to
better assess the variability of those estimates.</p>
<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> (0 minutes 10.868 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-auto-examples-ensemble-plot-gradient-boosting-quantile-py">
<div class="binder-badge docutils container">
<a class="reference external image-reference" href="https://mybinder.org/v2/gh/scikit-learn/scikit-learn/main?urlpath=lab/tree/notebooks/auto_examples/ensemble/plot_gradient_boosting_quantile.ipynb"><img alt="Launch binder" src="../../_images/binder_badge_logo11.svg" width="150px" /></a>
</div>
<div class="lite-badge docutils container">
<a class="reference external image-reference" href="../../lite/lab/?path=auto_examples/ensemble/plot_gradient_boosting_quantile.ipynb"><img alt="Launch JupyterLite" src="../../_images/jupyterlite_badge_logo11.svg" width="150px" /></a>
</div>
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/2f3ef774a6d7e52e1e6b7ccbb75d25f0/plot_gradient_boosting_quantile.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">plot_gradient_boosting_quantile.py</span></code></a></p>
</div>
<div class="sphx-glr-download sphx-glr-download-jupyter docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/b5ac5dfd67b0aab146fcb9faaac8480c/plot_gradient_boosting_quantile.ipynb"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Jupyter</span> <span class="pre">notebook:</span> <span class="pre">plot_gradient_boosting_quantile.ipynb</span></code></a></p>
</div>
</div>
<p class="sphx-glr-signature"><a class="reference external" href="https://sphinx-gallery.github.io">Gallery generated by Sphinx-Gallery</a></p>
</section>
</section>
</div>
<div class="container">
<footer class="sk-content-footer">
© 2007 - 2023, scikit-learn developers (BSD License).
<a href="../../_sources/auto_examples/ensemble/plot_gradient_boosting_quantile.rst.txt" rel="nofollow">Show this page source</a>
</footer>
</div>
</div>
</div>
<script src="../../_static/js/vendor/bootstrap.min.js"></script>
<script>
window.ga=window.ga||function(){(ga.q=ga.q||[]).push(arguments)};ga.l=+new Date;
ga('create', 'UA-22606712-2', 'auto');
ga('set', 'anonymizeIp', true);
ga('send', 'pageview');
</script>
<script async src='https://www.google-analytics.com/analytics.js'></script>
<script defer data-domain="scikit-learn.org" src="https://views.scientific-python.org/js/script.js">
</script>
<script src="../../_static/clipboard.min.js"></script>
<script src="../../_static/copybutton.js"></script>
<script>
$(document).ready(function() {
/* Add a [>>>] button on the top-right corner of code samples to hide
* the >>> and ... prompts and the output and thus make the code
* copyable. */
var div = $('.highlight-python .highlight,' +
'.highlight-python3 .highlight,' +
'.highlight-pycon .highlight,' +
'.highlight-default .highlight')
var pre = div.find('pre');
// get the styles from the current theme
pre.parent().parent().css('position', 'relative');
// create and add the button to all the code blocks that contain >>>
div.each(function(index) {
var jthis = $(this);
// tracebacks (.gt) contain bare text elements that need to be
// wrapped in a span to work with .nextUntil() (see later)
jthis.find('pre:has(.gt)').contents().filter(function() {
return ((this.nodeType == 3) && (this.data.trim().length > 0));
}).wrap('<span>');
});
/*** Add permalink buttons next to glossary terms ***/
$('dl.glossary > dt[id]').append(function() {
return ('<a class="headerlink" href="#' +
this.getAttribute('id') +
'" title="Permalink to this term">¶</a>');
});
});
</script>
<script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml.js"></script>
<script src="https://scikit-learn.org/versionwarning.js"></script>
</body>
</html>