-
Notifications
You must be signed in to change notification settings - Fork 81
/
Copy pathroadmap.html
543 lines (511 loc) · 32 KB
/
roadmap.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
<!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="Roadmap" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/roadmap.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="Purpose of this document: This document list general directions that core contributors are interested to see developed in scikit-learn. The fact that an item is listed here is in no way a promise t..." />
<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="Purpose of this document: This document list general directions that core contributors are interested to see developed in scikit-learn. The fact that an item is listed here is in no way a promise t..." />
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>Roadmap — scikit-learn 1.4.dev0 documentation</title>
<link rel="canonical" href="http://scikit-learn.org/stable/roadmap.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="auto_examples/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</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</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="whats_new/older_versions.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Version 0.12.1">Prev</a><a href="preface.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Welcome to scikit-learn">Up</a>
<a href="governance.html" role="button" class="btn sk-btn-rellink py-1" sk-rellink-tooltip="Scikit-learn governance and decision-making">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="#">Roadmap</a><ul>
<li><a class="reference internal" href="#purpose-of-this-document">Purpose of this document</a></li>
<li><a class="reference internal" href="#statement-of-purpose-scikit-learn-in-2018">Statement of purpose: Scikit-learn in 2018</a></li>
<li><a class="reference internal" href="#architectural-general-goals">Architectural / general goals</a></li>
<li><a class="reference internal" href="#subpackage-specific-goals">Subpackage-specific goals</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">
<span class="target" id="roadmap"></span><section id="id1">
<h1>Roadmap<a class="headerlink" href="#id1" title="Permalink to this heading">¶</a></h1>
<section id="purpose-of-this-document">
<h2>Purpose of this document<a class="headerlink" href="#purpose-of-this-document" title="Permalink to this heading">¶</a></h2>
<p>This document list general directions that core contributors are interested
to see developed in scikit-learn. The fact that an item is listed here is in
no way a promise that it will happen, as resources are limited. Rather, it
is an indication that help is welcomed on this topic.</p>
</section>
<section id="statement-of-purpose-scikit-learn-in-2018">
<h2>Statement of purpose: Scikit-learn in 2018<a class="headerlink" href="#statement-of-purpose-scikit-learn-in-2018" title="Permalink to this heading">¶</a></h2>
<p>Eleven years after the inception of Scikit-learn, much has changed in the
world of machine learning. Key changes include:</p>
<ul class="simple">
<li><p>Computational tools: The exploitation of GPUs, distributed programming
frameworks like Scala/Spark, etc.</p></li>
<li><p>High-level Python libraries for experimentation, processing and data
management: Jupyter notebook, Cython, Pandas, Dask, Numba…</p></li>
<li><p>Changes in the focus of machine learning research: artificial intelligence
applications (where input structure is key) with deep learning,
representation learning, reinforcement learning, domain transfer, etc.</p></li>
</ul>
<p>A more subtle change over the last decade is that, due to changing interests
in ML, PhD students in machine learning are more likely to contribute to
PyTorch, Dask, etc. than to Scikit-learn, so our contributor pool is very
different to a decade ago.</p>
<p>Scikit-learn remains very popular in practice for trying out canonical
machine learning techniques, particularly for applications in experimental
science and in data science. A lot of what we provide is now very mature.
But it can be costly to maintain, and we cannot therefore include arbitrary
new implementations. Yet Scikit-learn is also essential in defining an API
framework for the development of interoperable machine learning components
external to the core library.</p>
<p><strong>Thus our main goals in this era are to</strong>:</p>
<ul class="simple">
<li><p>continue maintaining a high-quality, well-documented collection of canonical
tools for data processing and machine learning within the current scope
(i.e. rectangular data largely invariant to column and row order;
predicting targets with simple structure)</p></li>
<li><p>improve the ease for users to develop and publish external components</p></li>
<li><p>improve interoperability with modern data science tools (e.g. Pandas, Dask)
and infrastructures (e.g. distributed processing)</p></li>
</ul>
<p>Many of the more fine-grained goals can be found under the <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3AAPI">API tag</a>
on the issue tracker.</p>
</section>
<section id="architectural-general-goals">
<h2>Architectural / general goals<a class="headerlink" href="#architectural-general-goals" title="Permalink to this heading">¶</a></h2>
<p>The list is numbered not as an indication of the order of priority, but to
make referring to specific points easier. Please add new entries only at the
bottom. Note that the crossed out entries are already done, and we try to keep
the document up to date as we work on these issues.</p>
<ol class="arabic simple">
<li><p>Improved handling of Pandas DataFrames</p>
<ul class="simple">
<li><p>document current handling</p></li>
<li><p>column reordering issue <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/7242">#7242</a></p></li>
<li><p>avoiding unnecessary conversion to ndarray <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/12147">#12147</a> </strike></p></li>
<li><p>returning DataFrames from transformers <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/5523">#5523</a></p></li>
<li><p>getting DataFrames from dataset loaders <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/10733">#10733</a> </strike>,
<strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13902">#13902</a> </strike></p></li>
<li><p>Sparse currently not considered <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/12800">#12800</a> </strike></p></li>
</ul>
</li>
<li><p>Improved handling of categorical features</p>
<ul class="simple">
<li><p>Tree-based models should be able to handle both continuous and categorical
features <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/12866">#12866</a> and <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/15550">#15550</a> </strike>.</p></li>
<li><p><strike> In dataset loaders <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13902">#13902</a> </strike></p></li>
<li><p>As generic transformers to be used with ColumnTransforms (e.g. ordinal
encoding supervised by correlation with target variable) <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/5853">#5853</a>,
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/11805">#11805</a></p></li>
<li><p>Handling mixtures of categorical and continuous variables</p></li>
</ul>
</li>
<li><p>Improved handling of missing data</p>
<ul class="simple">
<li><p>Making sure meta-estimators are lenient towards missing data,
<strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/15319">#15319</a> </strike></p></li>
<li><p>Non-trivial imputers <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/11977">#11977</a>, <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/12852">#12852</a> </strike></p></li>
<li><p>Learners directly handling missing data <strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13911">#13911</a> </strike></p></li>
<li><p>An amputation sample generator to make parts of a dataset go missing
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6284">#6284</a></p></li>
</ul>
</li>
<li><p>More didactic documentation</p>
<ul class="simple">
<li><p>More and more options have been added to scikit-learn. As a result, the
documentation is crowded which makes it hard for beginners to get the big
picture. Some work could be done in prioritizing the information.</p></li>
</ul>
</li>
<li><p>Passing around information that is not (X, y): Sample properties</p>
<ul class="simple">
<li><p>We need to be able to pass sample weights to scorers in cross validation.</p></li>
<li><p>We should have standard/generalised ways of passing sample-wise properties
around in meta-estimators. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/4497">#4497</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/7646">#7646</a></p></li>
</ul>
</li>
<li><p>Passing around information that is not (X, y): Feature properties</p>
<ul class="simple">
<li><p>Feature names or descriptions should ideally be available to fit for, e.g.
. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6425">#6425</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6424">#6424</a></p></li>
<li><p>Per-feature handling (e.g. “is this a nominal / ordinal / English language
text?”) should also not need to be provided to estimator constructors,
ideally, but should be available as metadata alongside X. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8480">#8480</a></p></li>
</ul>
</li>
<li><p>Passing around information that is not (X, y): Target information</p>
<ul class="simple">
<li><p>We have problems getting the full set of classes to all components when
the data is split/sampled. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6231">#6231</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8100">#8100</a></p></li>
<li><p>We have no way to handle a mixture of categorical and continuous targets.</p></li>
</ul>
</li>
<li><p>Make it easier for external users to write Scikit-learn-compatible
components</p>
<ul class="simple">
<li><p>More flexible estimator checks that do not select by estimator name
<strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6599">#6599</a> </strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6715">#6715</a></p></li>
<li><p>Example of how to develop an estimator or a meta-estimator,
<strike> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/14582">#14582</a> </strike></p></li>
<li><p>More self-sufficient running of scikit-learn-contrib or a similar resource</p></li>
</ul>
</li>
<li><p>Support resampling and sample reduction</p>
<ul class="simple">
<li><p>Allow subsampling of majority classes (in a pipeline?) <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/3855">#3855</a></p></li>
<li><p>Implement random forests with resampling <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13227">#13227</a></p></li>
</ul>
</li>
<li><p>Better interfaces for interactive development</p>
<ul class="simple">
<li><p><strike> __repr__ and HTML visualisations of estimators
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6323">#6323</a> and <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/pull/14180">#14180</a> </strike>.</p></li>
<li><p>Include plotting tools, not just as examples. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/9173">#9173</a></p></li>
</ul>
</li>
<li><p>Improved tools for model diagnostics and basic inference</p>
<ul class="simple">
<li><p><strike> alternative feature importances implementations, <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/13146">#13146</a> </strike></p></li>
<li><p>better ways to handle validation sets when fitting</p></li>
<li><p>better ways to find thresholds / create decision rules <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8614">#8614</a></p></li>
</ul>
</li>
<li><p>Better tools for selecting hyperparameters with transductive estimators</p>
<ul class="simple">
<li><p>Grid search and cross validation are not applicable to most clustering
tasks. Stability-based selection is more relevant.</p></li>
</ul>
</li>
<li><p>Better support for manual and automatic pipeline building</p>
<ul class="simple">
<li><p>Easier way to construct complex pipelines and valid search spaces
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/7608">#7608</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/5082">#5082</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8243">#8243</a></p></li>
<li><p>provide search ranges for common estimators??</p></li>
<li><p>cf. <a class="reference external" href="https://searchgrid.readthedocs.io/en/latest/">searchgrid</a></p></li>
</ul>
</li>
<li><p>Improved tracking of fitting</p>
<ul class="simple">
<li><p>Verbose is not very friendly and should use a standard logging library
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6929">#6929</a>, <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/78">#78</a></p></li>
<li><p>Callbacks or a similar system would facilitate logging and early stopping</p></li>
</ul>
</li>
<li><p>Distributed parallelism</p>
<ul class="simple">
<li><p>Accept data which complies with <code class="docutils literal notranslate"><span class="pre">__array_function__</span></code></p></li>
</ul>
</li>
<li><p>A way forward for more out of core</p>
<ul class="simple">
<li><p>Dask enables easy out-of-core computation. While the Dask model probably
cannot be adaptable to all machine-learning algorithms, most machine
learning is on smaller data than ETL, hence we can maybe adapt to very
large scale while supporting only a fraction of the patterns.</p></li>
</ul>
</li>
<li><p>Support for working with pre-trained models</p>
<ul class="simple">
<li><p>Estimator “freezing”. In particular, right now it’s impossible to clone a
<code class="docutils literal notranslate"><span class="pre">CalibratedClassifierCV</span></code> with prefit. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8370">#8370</a>. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/6451">#6451</a></p></li>
</ul>
</li>
<li><p>Backwards-compatible de/serialization of some estimators</p>
<ul class="simple">
<li><p>Currently serialization (with pickle) breaks across versions. While we may
not be able to get around other limitations of pickle re security etc, it
would be great to offer cross-version safety from version 1.0. Note: Gael
and Olivier think that this can cause heavy maintenance burden and we
should manage the trade-offs. A possible alternative is presented in the
following point.</p></li>
</ul>
</li>
<li><p>Documentation and tooling for model lifecycle management</p>
<ul class="simple">
<li><p>Document good practices for model deployments and lifecycle: before
deploying a model: snapshot the code versions (numpy, scipy, scikit-learn,
custom code repo), the training script and an alias on how to retrieve
historical training data + snapshot a copy of a small validation set +
snapshot of the predictions (predicted probabilities for classifiers)
on that validation set.</p></li>
<li><p>Document and tools to make it easy to manage upgrade of scikit-learn
versions:</p>
<ul>
<li><p>Try to load the old pickle, if it works, use the validation set
prediction snapshot to detect that the serialized model still behave
the same;</p></li>
<li><p>If joblib.load / pickle.load not work, use the versioned control
training script + historical training set to retrain the model and use
the validation set prediction snapshot to assert that it is possible to
recover the previous predictive performance: if this is not the case
there is probably a bug in scikit-learn that needs to be reported.</p></li>
</ul>
</li>
</ul>
</li>
<li><p>Everything in Scikit-learn should probably conform to our API contract.
We are still in the process of making decisions on some of these related
issues.</p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">Pipeline</span> <span class="pre"><pipeline.Pipeline></span></code> and <code class="docutils literal notranslate"><span class="pre">FeatureUnion</span></code> modify their input
parameters in fit. Fixing this requires making sure we have a good
grasp of their use cases to make sure all current functionality is
maintained. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/8157">#8157</a> <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/7382">#7382</a></p></li>
</ul>
</li>
<li><p>(Optional) Improve scikit-learn common tests suite to make sure that (at
least for frequently used) models have stable predictions across-versions
(to be discussed);</p>
<ul class="simple">
<li><p>Extend documentation to mention how to deploy models in Python-free
environments for instance <a class="reference external" href="https://github.com/onnx/sklearn-onnx">ONNX</a>.
and use the above best practices to assess predictive consistency between
scikit-learn and ONNX prediction functions on validation set.</p></li>
<li><p>Document good practices to detect temporal distribution drift for deployed
model and good practices for re-training on fresh data without causing
catastrophic predictive performance regressions.</p></li>
</ul>
</li>
</ol>
</section>
<section id="subpackage-specific-goals">
<h2>Subpackage-specific goals<a class="headerlink" href="#subpackage-specific-goals" title="Permalink to this heading">¶</a></h2>
<p><a class="reference internal" href="modules/classes.html#module-sklearn.ensemble" title="sklearn.ensemble"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.ensemble</span></code></a></p>
<ul class="simple">
<li><p><strike> a stacking implementation, <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/11047">#11047</a> </strike></p></li>
</ul>
<p><a class="reference internal" href="modules/classes.html#module-sklearn.cluster" title="sklearn.cluster"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.cluster</span></code></a></p>
<ul class="simple">
<li><p>kmeans variants for non-Euclidean distances, if we can show these have
benefits beyond hierarchical clustering.</p></li>
</ul>
<p><a class="reference internal" href="modules/classes.html#module-sklearn.model_selection" title="sklearn.model_selection"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.model_selection</span></code></a></p>
<ul class="simple">
<li><p><strike> multi-metric scoring is slow <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/9326">#9326</a> </strike></p></li>
<li><p>perhaps we want to be able to get back more than multiple metrics</p></li>
<li><p>the handling of random states in CV splitters is a poor design and
contradicts the validation of similar parameters in estimators,
<a class="reference external" href="https://github.com/scikit-learn/enhancement_proposals/pull/24">SLEP011</a></p></li>
<li><p>exploit warm-starting and path algorithms so the benefits of <code class="docutils literal notranslate"><span class="pre">EstimatorCV</span></code>
objects can be accessed via <code class="docutils literal notranslate"><span class="pre">GridSearchCV</span></code> and used in Pipelines.
<a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/1626">#1626</a></p></li>
<li><p>Cross-validation should be able to be replaced by OOB estimates whenever a
cross-validation iterator is used.</p></li>
<li><p>Redundant computations in pipelines should be avoided (related to point
above) cf <a class="reference external" href="https://ml.dask.org/hyper-parameter-search.html#avoid-repeated-work">dask-ml</a></p></li>
</ul>
<p><a class="reference internal" href="modules/classes.html#module-sklearn.neighbors" title="sklearn.neighbors"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.neighbors</span></code></a></p>
<ul class="simple">
<li><p><strike> Ability to substitute a custom/approximate/precomputed nearest neighbors
implementation for ours in all/most contexts that nearest neighbors are used
for learning. <a class="reference external" href="https://github.com/scikit-learn/scikit-learn/issues/10463">#10463</a> </strike></p></li>
</ul>
<p><a class="reference internal" href="modules/classes.html#module-sklearn.pipeline" title="sklearn.pipeline"><code class="xref py py-mod docutils literal notranslate"><span class="pre">sklearn.pipeline</span></code></a></p>
<ul class="simple">
<li><p>Performance issues with <code class="docutils literal notranslate"><span class="pre">Pipeline.memory</span></code></p></li>
<li><p>see “Everything in Scikit-learn should conform to our API contract” above</p></li>
</ul>
</section>
</section>
</div>
<div class="container">
<footer class="sk-content-footer">
© 2007 - 2023, scikit-learn developers (BSD License).
<a href="_sources/roadmap.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>