nirs4all-ui 0.1.16 → 0.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (138) hide show
  1. package/README.md +20 -0
  2. package/assets/dropzone.css +21 -0
  3. package/assets/pipeline-presets.css +38 -0
  4. package/assets/qualification.css +142 -0
  5. package/dist/components/DatasetDropzone.d.ts +44 -0
  6. package/dist/components/DatasetDropzone.d.ts.map +1 -0
  7. package/dist/components/DatasetDropzone.js +84 -0
  8. package/dist/components/DatasetDropzone.js.map +1 -0
  9. package/dist/components/PipelinePresetCard.d.ts +16 -0
  10. package/dist/components/PipelinePresetCard.d.ts.map +1 -0
  11. package/dist/components/PipelinePresetCard.js +21 -0
  12. package/dist/components/PipelinePresetCard.js.map +1 -0
  13. package/dist/components/index.d.ts +2 -0
  14. package/dist/components/index.d.ts.map +1 -1
  15. package/dist/components/index.js +2 -0
  16. package/dist/components/index.js.map +1 -1
  17. package/dist/index.d.ts +1 -0
  18. package/dist/index.d.ts.map +1 -1
  19. package/dist/index.js +1 -0
  20. package/dist/index.js.map +1 -1
  21. package/dist/pipelines/canonicalPipelinePreview.d.ts +21 -0
  22. package/dist/pipelines/canonicalPipelinePreview.d.ts.map +1 -0
  23. package/dist/pipelines/canonicalPipelinePreview.js +384 -0
  24. package/dist/pipelines/canonicalPipelinePreview.js.map +1 -0
  25. package/dist/pipelines/index.d.ts +5 -0
  26. package/dist/pipelines/index.d.ts.map +1 -0
  27. package/dist/pipelines/index.js +5 -0
  28. package/dist/pipelines/index.js.map +1 -0
  29. package/dist/pipelines/pipelineStats.d.ts +27 -0
  30. package/dist/pipelines/pipelineStats.d.ts.map +1 -0
  31. package/dist/pipelines/pipelineStats.js +128 -0
  32. package/dist/pipelines/pipelineStats.js.map +1 -0
  33. package/dist/pipelines/presets.d.ts +50 -0
  34. package/dist/pipelines/presets.d.ts.map +1 -0
  35. package/dist/pipelines/presets.js +92 -0
  36. package/dist/pipelines/presets.js.map +1 -0
  37. package/dist/pipelines/studio-presets.json +3996 -0
  38. package/dist/pipelines/studioPresets.d.ts +21 -0
  39. package/dist/pipelines/studioPresets.d.ts.map +1 -0
  40. package/dist/pipelines/studioPresets.js +12 -0
  41. package/dist/pipelines/studioPresets.js.map +1 -0
  42. package/dist/qualification/AvailabilitySummary.d.ts +3 -1
  43. package/dist/qualification/AvailabilitySummary.d.ts.map +1 -1
  44. package/dist/qualification/AvailabilitySummary.js +2 -2
  45. package/dist/qualification/AvailabilitySummary.js.map +1 -1
  46. package/dist/qualification/CapabilityGrid.d.ts +10 -0
  47. package/dist/qualification/CapabilityGrid.d.ts.map +1 -0
  48. package/dist/qualification/CapabilityGrid.js +16 -0
  49. package/dist/qualification/CapabilityGrid.js.map +1 -0
  50. package/dist/qualification/CheckList.d.ts +19 -0
  51. package/dist/qualification/CheckList.d.ts.map +1 -0
  52. package/dist/qualification/CheckList.js +14 -0
  53. package/dist/qualification/CheckList.js.map +1 -0
  54. package/dist/qualification/ComparisonMatrix.d.ts +37 -0
  55. package/dist/qualification/ComparisonMatrix.d.ts.map +1 -0
  56. package/dist/qualification/ComparisonMatrix.js +30 -0
  57. package/dist/qualification/ComparisonMatrix.js.map +1 -0
  58. package/dist/qualification/EigenvalueScree.d.ts +17 -0
  59. package/dist/qualification/EigenvalueScree.d.ts.map +1 -0
  60. package/dist/qualification/EigenvalueScree.js +39 -0
  61. package/dist/qualification/EigenvalueScree.js.map +1 -0
  62. package/dist/qualification/EvidenceTable.d.ts +3 -1
  63. package/dist/qualification/EvidenceTable.d.ts.map +1 -1
  64. package/dist/qualification/EvidenceTable.js +14 -6
  65. package/dist/qualification/EvidenceTable.js.map +1 -1
  66. package/dist/qualification/FindingList.d.ts +3 -1
  67. package/dist/qualification/FindingList.d.ts.map +1 -1
  68. package/dist/qualification/FindingList.js +5 -3
  69. package/dist/qualification/FindingList.js.map +1 -1
  70. package/dist/qualification/HypothesisTestTable.d.ts +48 -0
  71. package/dist/qualification/HypothesisTestTable.d.ts.map +1 -0
  72. package/dist/qualification/HypothesisTestTable.js +26 -0
  73. package/dist/qualification/HypothesisTestTable.js.map +1 -0
  74. package/dist/qualification/MeasureBars.d.ts +45 -0
  75. package/dist/qualification/MeasureBars.d.ts.map +1 -0
  76. package/dist/qualification/MeasureBars.js +52 -0
  77. package/dist/qualification/MeasureBars.js.map +1 -0
  78. package/dist/qualification/PopulationFlow.d.ts +42 -0
  79. package/dist/qualification/PopulationFlow.d.ts.map +1 -0
  80. package/dist/qualification/PopulationFlow.js +123 -0
  81. package/dist/qualification/PopulationFlow.js.map +1 -0
  82. package/dist/qualification/RepeatabilityProfile.d.ts +19 -0
  83. package/dist/qualification/RepeatabilityProfile.d.ts.map +1 -0
  84. package/dist/qualification/RepeatabilityProfile.js +119 -0
  85. package/dist/qualification/RepeatabilityProfile.js.map +1 -0
  86. package/dist/qualification/ReplicateHistogram.d.ts +23 -0
  87. package/dist/qualification/ReplicateHistogram.d.ts.map +1 -0
  88. package/dist/qualification/ReplicateHistogram.js +41 -0
  89. package/dist/qualification/ReplicateHistogram.js.map +1 -0
  90. package/dist/qualification/ScoreMap.d.ts +38 -0
  91. package/dist/qualification/ScoreMap.d.ts.map +1 -0
  92. package/dist/qualification/ScoreMap.js +77 -0
  93. package/dist/qualification/ScoreMap.js.map +1 -0
  94. package/dist/qualification/SpectraExplorer.d.ts +31 -0
  95. package/dist/qualification/SpectraExplorer.d.ts.map +1 -0
  96. package/dist/qualification/SpectraExplorer.js +111 -0
  97. package/dist/qualification/SpectraExplorer.js.map +1 -0
  98. package/dist/qualification/StatTile.d.ts +16 -0
  99. package/dist/qualification/StatTile.d.ts.map +1 -0
  100. package/dist/qualification/StatTile.js +10 -0
  101. package/dist/qualification/StatTile.js.map +1 -0
  102. package/dist/qualification/TargetDistribution.d.ts +35 -0
  103. package/dist/qualification/TargetDistribution.d.ts.map +1 -0
  104. package/dist/qualification/TargetDistribution.js +87 -0
  105. package/dist/qualification/TargetDistribution.js.map +1 -0
  106. package/dist/qualification/_chart.d.ts +19 -0
  107. package/dist/qualification/_chart.d.ts.map +1 -0
  108. package/dist/qualification/_chart.js +30 -0
  109. package/dist/qualification/_chart.js.map +1 -0
  110. package/dist/qualification/engineNumber.d.ts +26 -0
  111. package/dist/qualification/engineNumber.d.ts.map +1 -0
  112. package/dist/qualification/engineNumber.js +88 -0
  113. package/dist/qualification/engineNumber.js.map +1 -0
  114. package/dist/qualification/extract.d.ts +107 -0
  115. package/dist/qualification/extract.d.ts.map +1 -0
  116. package/dist/qualification/extract.js +139 -0
  117. package/dist/qualification/extract.js.map +1 -0
  118. package/dist/qualification/extractBlocks.d.ts +441 -0
  119. package/dist/qualification/extractBlocks.d.ts.map +1 -0
  120. package/dist/qualification/extractBlocks.js +576 -0
  121. package/dist/qualification/extractBlocks.js.map +1 -0
  122. package/dist/qualification/index.d.ts +16 -0
  123. package/dist/qualification/index.d.ts.map +1 -1
  124. package/dist/qualification/index.js +16 -0
  125. package/dist/qualification/index.js.map +1 -1
  126. package/dist/qualification/locale.d.ts +317 -1
  127. package/dist/qualification/locale.d.ts.map +1 -1
  128. package/dist/qualification/locale.js +369 -0
  129. package/dist/qualification/locale.js.map +1 -1
  130. package/dist/qualification/report.d.ts +13 -1
  131. package/dist/qualification/report.d.ts.map +1 -1
  132. package/dist/qualification/report.js +94 -3
  133. package/dist/qualification/report.js.map +1 -1
  134. package/dist/styles/index.d.ts +7 -1
  135. package/dist/styles/index.d.ts.map +1 -1
  136. package/dist/styles/index.js +5 -0
  137. package/dist/styles/index.js.map +1 -1
  138. package/package.json +7 -2
@@ -0,0 +1,3996 @@
1
+ {
2
+ "source": "nirs4all-studio/api/presets",
3
+ "commit": "2926c6a44b50715ee79a213ebbf523dba5d92f3d",
4
+ "files_sha256": {
5
+ "complex_pls.yaml": "d29397856ee3f32c27704b564fcd079ed8f78c03d91ec33b274fa454fff264f2",
6
+ "complex_trees.yaml": "ab5f88ec1c1f0adeff67e8e32d9a690389ec3b7fffeffc67bc38edbba231dccf",
7
+ "deep_nonlinear_exploration.yaml": "815f5ff1e960cb73cdec080f29ba1897983a0e37320f771cfd01f1f11f6c34e2",
8
+ "fast_result.yaml": "9b1a17e4cb3ab5bef0e9e06f0cfd29cfe4d07757417f98a3816656818a5ea074",
9
+ "nonlinear_exploration.yaml": "fb5bc52da3d05f9dd5de256213f398bca8d1375802cfe05d8928ab55d28a2844",
10
+ "simple_pls.yaml": "b78ccde2d0145809bfb49067aef71ff9a30abb5ddfe6e3f4e552cee05aefc8f4",
11
+ "simple_trees_boosting.yaml": "76f1c9948f66ecd07f7c2253fb460dc5e8f6d4acdf324a60662f463846a9958f",
12
+ "ultra_pls.yaml": "51c9c9ce88f47d813b4ff06b38b647e4e2b48be8796ae963f785022e9feedc65",
13
+ "ultra_slow.yaml": "e18f81d578c9e89c3e276c6c5f325b40fe1212aac20089a64a5ea0685cffa6f2",
14
+ "ultra_trees.yaml": "c0346f612ea050ad97f1097e1c1e6b336dcd426e84bb79f9271e1745e061227d"
15
+ },
16
+ "presets": [
17
+ {
18
+ "id": "simple_pls",
19
+ "name": "Simple PLS",
20
+ "description": "Three classic preprocessing choices, KFold, and a single n_components sweep for PLS.",
21
+ "complexity": 1,
22
+ "default_variant": "regression",
23
+ "variants": {
24
+ "regression": {
25
+ "format": "yaml",
26
+ "pipeline": [
27
+ {
28
+ "_or_": [
29
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
30
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
31
+ {
32
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
33
+ "params": {
34
+ "window_length": 15,
35
+ "polyorder": 2,
36
+ "deriv": 1
37
+ }
38
+ }
39
+ ]
40
+ },
41
+ {
42
+ "class": "sklearn.model_selection._split.KFold",
43
+ "params": {
44
+ "shuffle": true,
45
+ "random_state": 42
46
+ }
47
+ },
48
+ {
49
+ "model": "sklearn.cross_decomposition._pls.PLSRegression",
50
+ "name": "PLS-Sweep",
51
+ "_range_": [
52
+ 2,
53
+ 30,
54
+ 4
55
+ ],
56
+ "param": "n_components"
57
+ }
58
+ ]
59
+ },
60
+ "classification": {
61
+ "format": "yaml",
62
+ "pipeline": [
63
+ {
64
+ "_or_": [
65
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
66
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
67
+ {
68
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
69
+ "params": {
70
+ "window_length": 15,
71
+ "polyorder": 2,
72
+ "deriv": 1
73
+ }
74
+ }
75
+ ]
76
+ },
77
+ {
78
+ "class": "sklearn.model_selection._split.StratifiedKFold",
79
+ "params": {
80
+ "shuffle": true,
81
+ "random_state": 42
82
+ }
83
+ },
84
+ {
85
+ "model": "nirs4all.operators.models.sklearn.plsda.PLSDA",
86
+ "name": "PLSDA-Sweep",
87
+ "_range_": [
88
+ 2,
89
+ 30,
90
+ 4
91
+ ],
92
+ "param": "n_components"
93
+ }
94
+ ]
95
+ }
96
+ },
97
+ "available_variants": [
98
+ "regression",
99
+ "classification"
100
+ ],
101
+ "steps_count": 3
102
+ },
103
+ {
104
+ "id": "fast_result",
105
+ "name": "Fast Result",
106
+ "description": "Quick SPXY-oriented preset with simple preprocessing, AOM-PLS, and TabPFN for fast first results.",
107
+ "complexity": 2,
108
+ "default_variant": "regression",
109
+ "variants": {
110
+ "regression": {
111
+ "format": "yaml",
112
+ "pipeline": [
113
+ {
114
+ "_or_": [
115
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
116
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
117
+ {
118
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
119
+ "params": {
120
+ "window_length": 15,
121
+ "polyorder": 2,
122
+ "deriv": 1
123
+ }
124
+ }
125
+ ]
126
+ },
127
+ {
128
+ "class": "nirs4all.operators.splitters.splitters.SPXYFold",
129
+ "params": {
130
+ "n_splits": 3,
131
+ "random_state": 42
132
+ }
133
+ },
134
+ {
135
+ "model": {
136
+ "class": "nirs4all.operators.models.sklearn.aom_pls.AOMPLSRegressor",
137
+ "params": {
138
+ "n_components": 12,
139
+ "center": true,
140
+ "scale": false
141
+ }
142
+ },
143
+ "name": "AOM-PLS"
144
+ },
145
+ {
146
+ "model": {
147
+ "class": "tabpfn.TabPFNRegressor",
148
+ "params": {
149
+ "ignore_pretraining_limits": true
150
+ }
151
+ },
152
+ "name": "TabPFN"
153
+ }
154
+ ]
155
+ },
156
+ "classification": {
157
+ "format": "yaml",
158
+ "pipeline": [
159
+ {
160
+ "_or_": [
161
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
162
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
163
+ {
164
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
165
+ "params": {
166
+ "window_length": 15,
167
+ "polyorder": 2,
168
+ "deriv": 1
169
+ }
170
+ }
171
+ ]
172
+ },
173
+ {
174
+ "class": "sklearn.model_selection._split.StratifiedKFold",
175
+ "params": {
176
+ "n_splits": 3,
177
+ "shuffle": true,
178
+ "random_state": 42
179
+ }
180
+ },
181
+ {
182
+ "model": {
183
+ "class": "nirs4all.operators.models.sklearn.aom_pls_classifier.AOMPLSClassifier",
184
+ "params": {
185
+ "n_components": 12
186
+ }
187
+ },
188
+ "name": "AOM-PLS"
189
+ },
190
+ {
191
+ "model": {
192
+ "class": "tabpfn.TabPFNClassifier",
193
+ "params": {
194
+ "ignore_pretraining_limits": true
195
+ }
196
+ },
197
+ "name": "TabPFN"
198
+ }
199
+ ]
200
+ }
201
+ },
202
+ "available_variants": [
203
+ "regression",
204
+ "classification"
205
+ ],
206
+ "steps_count": 4
207
+ },
208
+ {
209
+ "id": "simple_trees_boosting",
210
+ "name": "Simple Trees / Boosting",
211
+ "description": "Same lightweight preprocessing pattern as Simple PLS, with LightGBM and RandomForest sweeps.",
212
+ "complexity": 3,
213
+ "default_variant": "regression",
214
+ "variants": {
215
+ "regression": {
216
+ "format": "yaml",
217
+ "pipeline": [
218
+ {
219
+ "_or_": [
220
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
221
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
222
+ {
223
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
224
+ "params": {
225
+ "window_length": 15,
226
+ "polyorder": 2,
227
+ "deriv": 1
228
+ }
229
+ }
230
+ ]
231
+ },
232
+ {
233
+ "class": "sklearn.model_selection._split.KFold",
234
+ "params": {
235
+ "shuffle": true,
236
+ "random_state": 42
237
+ }
238
+ },
239
+ {
240
+ "model": {
241
+ "class": "lightgbm.LGBMRegressor",
242
+ "params": {
243
+ "random_state": 42,
244
+ "verbose": -1
245
+ }
246
+ },
247
+ "name": "LightGBM-Sweep",
248
+ "finetune_params": {
249
+ "n_trials": 6,
250
+ "sampler": "grid",
251
+ "approach": "single",
252
+ "eval_mode": "best",
253
+ "verbose": 0,
254
+ "model_params": {
255
+ "n_estimators": [
256
+ 100,
257
+ 200,
258
+ 400
259
+ ],
260
+ "max_depth": [
261
+ 6,
262
+ 10
263
+ ],
264
+ "learning_rate": [
265
+ 0.05
266
+ ]
267
+ }
268
+ }
269
+ },
270
+ {
271
+ "model": {
272
+ "class": "sklearn.ensemble._forest.RandomForestRegressor",
273
+ "params": {
274
+ "random_state": 42
275
+ }
276
+ },
277
+ "name": "RandomForest-Sweep",
278
+ "finetune_params": {
279
+ "n_trials": 6,
280
+ "sampler": "grid",
281
+ "approach": "single",
282
+ "eval_mode": "best",
283
+ "verbose": 0,
284
+ "model_params": {
285
+ "n_estimators": [
286
+ 200,
287
+ 400,
288
+ 800
289
+ ],
290
+ "max_depth": [
291
+ 8,
292
+ 16
293
+ ]
294
+ }
295
+ }
296
+ }
297
+ ]
298
+ },
299
+ "classification": {
300
+ "format": "yaml",
301
+ "pipeline": [
302
+ {
303
+ "_or_": [
304
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
305
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
306
+ {
307
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
308
+ "params": {
309
+ "window_length": 15,
310
+ "polyorder": 2,
311
+ "deriv": 1
312
+ }
313
+ }
314
+ ]
315
+ },
316
+ {
317
+ "class": "sklearn.model_selection._split.StratifiedKFold",
318
+ "params": {
319
+ "shuffle": true,
320
+ "random_state": 42
321
+ }
322
+ },
323
+ {
324
+ "model": {
325
+ "class": "lightgbm.LGBMClassifier",
326
+ "params": {
327
+ "random_state": 42,
328
+ "verbose": -1
329
+ }
330
+ },
331
+ "name": "LightGBM-Sweep",
332
+ "finetune_params": {
333
+ "n_trials": 6,
334
+ "sampler": "grid",
335
+ "approach": "single",
336
+ "eval_mode": "best",
337
+ "verbose": 0,
338
+ "model_params": {
339
+ "n_estimators": [
340
+ 100,
341
+ 200,
342
+ 400
343
+ ],
344
+ "max_depth": [
345
+ 6,
346
+ 10
347
+ ],
348
+ "learning_rate": [
349
+ 0.05
350
+ ]
351
+ }
352
+ }
353
+ },
354
+ {
355
+ "model": {
356
+ "class": "sklearn.ensemble._forest.RandomForestClassifier",
357
+ "params": {
358
+ "random_state": 42
359
+ }
360
+ },
361
+ "name": "RandomForest-Sweep",
362
+ "finetune_params": {
363
+ "n_trials": 6,
364
+ "sampler": "grid",
365
+ "approach": "single",
366
+ "eval_mode": "best",
367
+ "verbose": 0,
368
+ "model_params": {
369
+ "n_estimators": [
370
+ 200,
371
+ 400,
372
+ 800
373
+ ],
374
+ "max_depth": [
375
+ 8,
376
+ 16
377
+ ]
378
+ }
379
+ }
380
+ }
381
+ ]
382
+ }
383
+ },
384
+ "available_variants": [
385
+ "regression",
386
+ "classification"
387
+ ],
388
+ "steps_count": 4
389
+ },
390
+ {
391
+ "id": "complex_pls",
392
+ "name": "Complex PLS",
393
+ "description": "Benchmark-style cartesian preprocessing with fixed PLS, tuned PLS, tuned Ridge, and AOM-PLS.",
394
+ "complexity": 4,
395
+ "default_variant": "regression",
396
+ "variants": {
397
+ "regression": {
398
+ "format": "yaml",
399
+ "pipeline": [
400
+ {
401
+ "_cartesian_": [
402
+ {
403
+ "_or_": [
404
+ null,
405
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
406
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
407
+ {
408
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
409
+ "params": {
410
+ "degree": 1
411
+ }
412
+ },
413
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
414
+ ]
415
+ },
416
+ {
417
+ "_or_": [
418
+ null,
419
+ {
420
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
421
+ "params": {
422
+ "polyorder": 2,
423
+ "deriv": 1
424
+ }
425
+ },
426
+ {
427
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
428
+ "params": {
429
+ "window_length": 15,
430
+ "polyorder": 2,
431
+ "deriv": 1
432
+ }
433
+ },
434
+ {
435
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
436
+ "params": {
437
+ "window_length": 21,
438
+ "polyorder": 2,
439
+ "deriv": 1
440
+ }
441
+ },
442
+ {
443
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
444
+ "params": {
445
+ "window_length": 31,
446
+ "polyorder": 2,
447
+ "deriv": 1
448
+ }
449
+ },
450
+ {
451
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
452
+ "params": {
453
+ "window_length": 15,
454
+ "deriv": 2
455
+ }
456
+ },
457
+ {
458
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
459
+ "params": {
460
+ "window_length": 21,
461
+ "deriv": 2
462
+ }
463
+ },
464
+ {
465
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
466
+ "params": {
467
+ "window_length": 31,
468
+ "deriv": 2
469
+ }
470
+ },
471
+ {
472
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
473
+ "params": {
474
+ "order": 0
475
+ }
476
+ },
477
+ {
478
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
479
+ "params": {
480
+ "order": 0,
481
+ "sigma": 2
482
+ }
483
+ }
484
+ ]
485
+ },
486
+ {
487
+ "_or_": [
488
+ null,
489
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
490
+ "nirs4all.operators.transforms.signal.Detrend"
491
+ ]
492
+ },
493
+ {
494
+ "_or_": [
495
+ null,
496
+ "nirs4all.operators.transforms.orthogonalization.OSC",
497
+ {
498
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
499
+ "params": {
500
+ "n_components": 2
501
+ }
502
+ },
503
+ {
504
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
505
+ "params": {
506
+ "n_components": 3
507
+ }
508
+ }
509
+ ]
510
+ }
511
+ ],
512
+ "count": 150
513
+ },
514
+ {
515
+ "class": "sklearn.model_selection._split.KFold",
516
+ "params": {
517
+ "n_splits": 3,
518
+ "shuffle": true,
519
+ "random_state": 42
520
+ }
521
+ },
522
+ {
523
+ "class": "sklearn.preprocessing._data.StandardScaler",
524
+ "params": {
525
+ "with_std": false
526
+ }
527
+ },
528
+ {
529
+ "model": {
530
+ "class": "sklearn.cross_decomposition._pls.PLSRegression",
531
+ "params": {
532
+ "n_components": 12,
533
+ "scale": false
534
+ }
535
+ },
536
+ "name": "PLS-12"
537
+ },
538
+ {
539
+ "model": {
540
+ "class": "sklearn.cross_decomposition._pls.PLSRegression",
541
+ "params": {
542
+ "scale": false
543
+ }
544
+ },
545
+ "name": "PLS-Finetuned",
546
+ "finetune_params": {
547
+ "n_trials": 25,
548
+ "sampler": "binary",
549
+ "model_params": {
550
+ "n_components": [
551
+ "int",
552
+ 1,
553
+ 25
554
+ ]
555
+ }
556
+ }
557
+ },
558
+ {
559
+ "model": "sklearn.linear_model._ridge.Ridge",
560
+ "name": "Ridge-Finetuned",
561
+ "finetune_params": {
562
+ "n_trials": 40,
563
+ "sampler": "tpe",
564
+ "model_params": {
565
+ "alpha": [
566
+ "log_float",
567
+ 1e-05,
568
+ 10000.0
569
+ ],
570
+ "fit_intercept": [
571
+ "categorical",
572
+ [
573
+ true,
574
+ false
575
+ ]
576
+ ],
577
+ "solver": [
578
+ "categorical",
579
+ [
580
+ "auto",
581
+ "svd",
582
+ "cholesky",
583
+ "lsqr"
584
+ ]
585
+ ]
586
+ }
587
+ }
588
+ },
589
+ {
590
+ "model": {
591
+ "class": "nirs4all.operators.models.sklearn.aom_pls.AOMPLSRegressor",
592
+ "params": {
593
+ "n_components": 12,
594
+ "center": true,
595
+ "scale": false
596
+ }
597
+ },
598
+ "name": "AOM-PLS"
599
+ }
600
+ ]
601
+ },
602
+ "classification": {
603
+ "format": "yaml",
604
+ "pipeline": [
605
+ {
606
+ "_cartesian_": [
607
+ {
608
+ "_or_": [
609
+ null,
610
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
611
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
612
+ {
613
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
614
+ "params": {
615
+ "degree": 1
616
+ }
617
+ },
618
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
619
+ ]
620
+ },
621
+ {
622
+ "_or_": [
623
+ null,
624
+ {
625
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
626
+ "params": {
627
+ "polyorder": 2,
628
+ "deriv": 1
629
+ }
630
+ },
631
+ {
632
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
633
+ "params": {
634
+ "window_length": 15,
635
+ "polyorder": 2,
636
+ "deriv": 1
637
+ }
638
+ },
639
+ {
640
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
641
+ "params": {
642
+ "window_length": 21,
643
+ "polyorder": 2,
644
+ "deriv": 1
645
+ }
646
+ },
647
+ {
648
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
649
+ "params": {
650
+ "window_length": 31,
651
+ "polyorder": 2,
652
+ "deriv": 1
653
+ }
654
+ },
655
+ {
656
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
657
+ "params": {
658
+ "window_length": 15,
659
+ "deriv": 2
660
+ }
661
+ },
662
+ {
663
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
664
+ "params": {
665
+ "window_length": 21,
666
+ "deriv": 2
667
+ }
668
+ },
669
+ {
670
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
671
+ "params": {
672
+ "window_length": 31,
673
+ "deriv": 2
674
+ }
675
+ },
676
+ {
677
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
678
+ "params": {
679
+ "order": 0
680
+ }
681
+ },
682
+ {
683
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
684
+ "params": {
685
+ "order": 0,
686
+ "sigma": 2
687
+ }
688
+ }
689
+ ]
690
+ },
691
+ {
692
+ "_or_": [
693
+ null,
694
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
695
+ "nirs4all.operators.transforms.signal.Detrend"
696
+ ]
697
+ },
698
+ {
699
+ "_or_": [
700
+ null,
701
+ "nirs4all.operators.transforms.orthogonalization.OSC",
702
+ {
703
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
704
+ "params": {
705
+ "n_components": 2
706
+ }
707
+ },
708
+ {
709
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
710
+ "params": {
711
+ "n_components": 3
712
+ }
713
+ }
714
+ ]
715
+ }
716
+ ],
717
+ "count": 150
718
+ },
719
+ {
720
+ "class": "sklearn.model_selection._split.StratifiedKFold",
721
+ "params": {
722
+ "n_splits": 3,
723
+ "shuffle": true,
724
+ "random_state": 42
725
+ }
726
+ },
727
+ {
728
+ "class": "sklearn.preprocessing._data.StandardScaler",
729
+ "params": {
730
+ "with_std": false
731
+ }
732
+ },
733
+ {
734
+ "model": {
735
+ "class": "nirs4all.operators.models.sklearn.plsda.PLSDA",
736
+ "params": {
737
+ "n_components": 12
738
+ }
739
+ },
740
+ "name": "PLSDA-12"
741
+ },
742
+ {
743
+ "model": "nirs4all.operators.models.sklearn.plsda.PLSDA",
744
+ "name": "PLSDA-Finetuned",
745
+ "finetune_params": {
746
+ "n_trials": 25,
747
+ "sampler": "binary",
748
+ "model_params": {
749
+ "n_components": [
750
+ "int",
751
+ 1,
752
+ 25
753
+ ]
754
+ }
755
+ }
756
+ },
757
+ {
758
+ "model": "sklearn.linear_model._ridge.RidgeClassifier",
759
+ "name": "RidgeClassifier-Finetuned",
760
+ "finetune_params": {
761
+ "n_trials": 30,
762
+ "sampler": "tpe",
763
+ "model_params": {
764
+ "alpha": [
765
+ "log_float",
766
+ 1e-05,
767
+ 10000.0
768
+ ],
769
+ "fit_intercept": [
770
+ "categorical",
771
+ [
772
+ true,
773
+ false
774
+ ]
775
+ ]
776
+ }
777
+ }
778
+ },
779
+ {
780
+ "model": {
781
+ "class": "nirs4all.operators.models.sklearn.aom_pls_classifier.AOMPLSClassifier",
782
+ "params": {
783
+ "n_components": 12
784
+ }
785
+ },
786
+ "name": "AOM-PLS"
787
+ }
788
+ ]
789
+ }
790
+ },
791
+ "available_variants": [
792
+ "regression",
793
+ "classification"
794
+ ],
795
+ "steps_count": 7
796
+ },
797
+ {
798
+ "id": "complex_trees",
799
+ "name": "Complex Trees",
800
+ "description": "Cartesian preprocessing with LightGBM, RandomForest, and XGBoost tree/boosting grids.",
801
+ "complexity": 5,
802
+ "default_variant": "regression",
803
+ "variants": {
804
+ "regression": {
805
+ "format": "yaml",
806
+ "pipeline": [
807
+ {
808
+ "_cartesian_": [
809
+ {
810
+ "_or_": [
811
+ null,
812
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
813
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
814
+ {
815
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
816
+ "params": {
817
+ "degree": 1
818
+ }
819
+ },
820
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
821
+ ]
822
+ },
823
+ {
824
+ "_or_": [
825
+ null,
826
+ {
827
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
828
+ "params": {
829
+ "polyorder": 2,
830
+ "deriv": 1
831
+ }
832
+ },
833
+ {
834
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
835
+ "params": {
836
+ "window_length": 15,
837
+ "polyorder": 2,
838
+ "deriv": 1
839
+ }
840
+ },
841
+ {
842
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
843
+ "params": {
844
+ "window_length": 21,
845
+ "polyorder": 2,
846
+ "deriv": 1
847
+ }
848
+ },
849
+ {
850
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
851
+ "params": {
852
+ "window_length": 31,
853
+ "polyorder": 2,
854
+ "deriv": 1
855
+ }
856
+ },
857
+ {
858
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
859
+ "params": {
860
+ "window_length": 15,
861
+ "deriv": 2
862
+ }
863
+ },
864
+ {
865
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
866
+ "params": {
867
+ "window_length": 21,
868
+ "deriv": 2
869
+ }
870
+ },
871
+ {
872
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
873
+ "params": {
874
+ "window_length": 31,
875
+ "deriv": 2
876
+ }
877
+ },
878
+ {
879
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
880
+ "params": {
881
+ "order": 0
882
+ }
883
+ },
884
+ {
885
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
886
+ "params": {
887
+ "order": 0,
888
+ "sigma": 2
889
+ }
890
+ }
891
+ ]
892
+ },
893
+ {
894
+ "_or_": [
895
+ null,
896
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
897
+ "nirs4all.operators.transforms.signal.Detrend"
898
+ ]
899
+ },
900
+ {
901
+ "_or_": [
902
+ null,
903
+ "nirs4all.operators.transforms.orthogonalization.OSC",
904
+ {
905
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
906
+ "params": {
907
+ "n_components": 2
908
+ }
909
+ },
910
+ {
911
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
912
+ "params": {
913
+ "n_components": 3
914
+ }
915
+ }
916
+ ]
917
+ }
918
+ ],
919
+ "count": 40
920
+ },
921
+ {
922
+ "class": "sklearn.model_selection._split.KFold",
923
+ "params": {
924
+ "n_splits": 3,
925
+ "shuffle": true,
926
+ "random_state": 42
927
+ }
928
+ },
929
+ {
930
+ "model": {
931
+ "class": "lightgbm.LGBMRegressor",
932
+ "params": {
933
+ "random_state": 42,
934
+ "verbose": -1
935
+ }
936
+ },
937
+ "name": "LightGBM-Sweep",
938
+ "finetune_params": {
939
+ "n_trials": 12,
940
+ "sampler": "grid",
941
+ "approach": "single",
942
+ "eval_mode": "best",
943
+ "verbose": 0,
944
+ "model_params": {
945
+ "n_estimators": [
946
+ 150,
947
+ 300,
948
+ 600
949
+ ],
950
+ "max_depth": [
951
+ 6,
952
+ 10
953
+ ],
954
+ "learning_rate": [
955
+ 0.03,
956
+ 0.05
957
+ ]
958
+ }
959
+ }
960
+ },
961
+ {
962
+ "model": {
963
+ "class": "sklearn.ensemble._forest.RandomForestRegressor",
964
+ "params": {
965
+ "random_state": 42
966
+ }
967
+ },
968
+ "name": "RandomForest-Sweep",
969
+ "finetune_params": {
970
+ "n_trials": 12,
971
+ "sampler": "grid",
972
+ "approach": "single",
973
+ "eval_mode": "best",
974
+ "verbose": 0,
975
+ "model_params": {
976
+ "n_estimators": [
977
+ 300,
978
+ 600
979
+ ],
980
+ "max_depth": [
981
+ 8,
982
+ 12,
983
+ 20
984
+ ],
985
+ "min_samples_leaf": [
986
+ 1,
987
+ 3
988
+ ]
989
+ }
990
+ }
991
+ },
992
+ {
993
+ "model": {
994
+ "class": "xgboost.XGBRegressor",
995
+ "params": {
996
+ "random_state": 42,
997
+ "verbosity": 0
998
+ }
999
+ },
1000
+ "name": "XGBoost-Sweep",
1001
+ "finetune_params": {
1002
+ "n_trials": 18,
1003
+ "sampler": "grid",
1004
+ "approach": "single",
1005
+ "eval_mode": "best",
1006
+ "verbose": 0,
1007
+ "model_params": {
1008
+ "n_estimators": [
1009
+ 150,
1010
+ 300,
1011
+ 600
1012
+ ],
1013
+ "max_depth": [
1014
+ 4,
1015
+ 6,
1016
+ 8
1017
+ ],
1018
+ "learning_rate": [
1019
+ 0.03,
1020
+ 0.05
1021
+ ]
1022
+ }
1023
+ }
1024
+ }
1025
+ ]
1026
+ },
1027
+ "classification": {
1028
+ "format": "yaml",
1029
+ "pipeline": [
1030
+ {
1031
+ "_cartesian_": [
1032
+ {
1033
+ "_or_": [
1034
+ null,
1035
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
1036
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
1037
+ {
1038
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
1039
+ "params": {
1040
+ "degree": 1
1041
+ }
1042
+ },
1043
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
1044
+ ]
1045
+ },
1046
+ {
1047
+ "_or_": [
1048
+ null,
1049
+ {
1050
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1051
+ "params": {
1052
+ "polyorder": 2,
1053
+ "deriv": 1
1054
+ }
1055
+ },
1056
+ {
1057
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1058
+ "params": {
1059
+ "window_length": 15,
1060
+ "polyorder": 2,
1061
+ "deriv": 1
1062
+ }
1063
+ },
1064
+ {
1065
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1066
+ "params": {
1067
+ "window_length": 21,
1068
+ "polyorder": 2,
1069
+ "deriv": 1
1070
+ }
1071
+ },
1072
+ {
1073
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1074
+ "params": {
1075
+ "window_length": 31,
1076
+ "polyorder": 2,
1077
+ "deriv": 1
1078
+ }
1079
+ },
1080
+ {
1081
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1082
+ "params": {
1083
+ "window_length": 15,
1084
+ "deriv": 2
1085
+ }
1086
+ },
1087
+ {
1088
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1089
+ "params": {
1090
+ "window_length": 21,
1091
+ "deriv": 2
1092
+ }
1093
+ },
1094
+ {
1095
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1096
+ "params": {
1097
+ "window_length": 31,
1098
+ "deriv": 2
1099
+ }
1100
+ },
1101
+ {
1102
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1103
+ "params": {
1104
+ "order": 0
1105
+ }
1106
+ },
1107
+ {
1108
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1109
+ "params": {
1110
+ "order": 0,
1111
+ "sigma": 2
1112
+ }
1113
+ }
1114
+ ]
1115
+ },
1116
+ {
1117
+ "_or_": [
1118
+ null,
1119
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
1120
+ "nirs4all.operators.transforms.signal.Detrend"
1121
+ ]
1122
+ },
1123
+ {
1124
+ "_or_": [
1125
+ null,
1126
+ "nirs4all.operators.transforms.orthogonalization.OSC",
1127
+ {
1128
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1129
+ "params": {
1130
+ "n_components": 2
1131
+ }
1132
+ },
1133
+ {
1134
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1135
+ "params": {
1136
+ "n_components": 3
1137
+ }
1138
+ }
1139
+ ]
1140
+ }
1141
+ ],
1142
+ "count": 40
1143
+ },
1144
+ {
1145
+ "class": "sklearn.model_selection._split.StratifiedKFold",
1146
+ "params": {
1147
+ "n_splits": 3,
1148
+ "shuffle": true,
1149
+ "random_state": 42
1150
+ }
1151
+ },
1152
+ {
1153
+ "model": {
1154
+ "class": "lightgbm.LGBMClassifier",
1155
+ "params": {
1156
+ "random_state": 42,
1157
+ "verbose": -1
1158
+ }
1159
+ },
1160
+ "name": "LightGBM-Sweep",
1161
+ "finetune_params": {
1162
+ "n_trials": 12,
1163
+ "sampler": "grid",
1164
+ "approach": "single",
1165
+ "eval_mode": "best",
1166
+ "verbose": 0,
1167
+ "model_params": {
1168
+ "n_estimators": [
1169
+ 150,
1170
+ 300,
1171
+ 600
1172
+ ],
1173
+ "max_depth": [
1174
+ 6,
1175
+ 10
1176
+ ],
1177
+ "learning_rate": [
1178
+ 0.03,
1179
+ 0.05
1180
+ ]
1181
+ }
1182
+ }
1183
+ },
1184
+ {
1185
+ "model": {
1186
+ "class": "sklearn.ensemble._forest.RandomForestClassifier",
1187
+ "params": {
1188
+ "random_state": 42
1189
+ }
1190
+ },
1191
+ "name": "RandomForest-Sweep",
1192
+ "finetune_params": {
1193
+ "n_trials": 12,
1194
+ "sampler": "grid",
1195
+ "approach": "single",
1196
+ "eval_mode": "best",
1197
+ "verbose": 0,
1198
+ "model_params": {
1199
+ "n_estimators": [
1200
+ 300,
1201
+ 600
1202
+ ],
1203
+ "max_depth": [
1204
+ 8,
1205
+ 12,
1206
+ 20
1207
+ ],
1208
+ "min_samples_leaf": [
1209
+ 1,
1210
+ 3
1211
+ ]
1212
+ }
1213
+ }
1214
+ },
1215
+ {
1216
+ "model": {
1217
+ "class": "xgboost.XGBClassifier",
1218
+ "params": {
1219
+ "random_state": 42,
1220
+ "verbosity": 0
1221
+ }
1222
+ },
1223
+ "name": "XGBoost-Sweep",
1224
+ "finetune_params": {
1225
+ "n_trials": 18,
1226
+ "sampler": "grid",
1227
+ "approach": "single",
1228
+ "eval_mode": "best",
1229
+ "verbose": 0,
1230
+ "model_params": {
1231
+ "n_estimators": [
1232
+ 150,
1233
+ 300,
1234
+ 600
1235
+ ],
1236
+ "max_depth": [
1237
+ 4,
1238
+ 6,
1239
+ 8
1240
+ ],
1241
+ "learning_rate": [
1242
+ 0.03,
1243
+ 0.05
1244
+ ]
1245
+ }
1246
+ }
1247
+ }
1248
+ ]
1249
+ }
1250
+ },
1251
+ "available_variants": [
1252
+ "regression",
1253
+ "classification"
1254
+ ],
1255
+ "steps_count": 5
1256
+ },
1257
+ {
1258
+ "id": "nonlinear_exploration",
1259
+ "name": "Non-Linear Exploration",
1260
+ "description": "Cartesian preprocessing with LightGBM, XGBoost, NICON, and TabPFN for broad non-linear screening.",
1261
+ "complexity": 6,
1262
+ "default_variant": "regression",
1263
+ "variants": {
1264
+ "regression": {
1265
+ "format": "yaml",
1266
+ "pipeline": [
1267
+ {
1268
+ "_cartesian_": [
1269
+ {
1270
+ "_or_": [
1271
+ null,
1272
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
1273
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
1274
+ {
1275
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
1276
+ "params": {
1277
+ "degree": 1
1278
+ }
1279
+ },
1280
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
1281
+ ]
1282
+ },
1283
+ {
1284
+ "_or_": [
1285
+ null,
1286
+ {
1287
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1288
+ "params": {
1289
+ "polyorder": 2,
1290
+ "deriv": 1
1291
+ }
1292
+ },
1293
+ {
1294
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1295
+ "params": {
1296
+ "window_length": 15,
1297
+ "polyorder": 2,
1298
+ "deriv": 1
1299
+ }
1300
+ },
1301
+ {
1302
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1303
+ "params": {
1304
+ "window_length": 21,
1305
+ "polyorder": 2,
1306
+ "deriv": 1
1307
+ }
1308
+ },
1309
+ {
1310
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1311
+ "params": {
1312
+ "window_length": 31,
1313
+ "polyorder": 2,
1314
+ "deriv": 1
1315
+ }
1316
+ },
1317
+ {
1318
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1319
+ "params": {
1320
+ "window_length": 15,
1321
+ "deriv": 2
1322
+ }
1323
+ },
1324
+ {
1325
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1326
+ "params": {
1327
+ "window_length": 21,
1328
+ "deriv": 2
1329
+ }
1330
+ },
1331
+ {
1332
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1333
+ "params": {
1334
+ "window_length": 31,
1335
+ "deriv": 2
1336
+ }
1337
+ },
1338
+ {
1339
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1340
+ "params": {
1341
+ "order": 0
1342
+ }
1343
+ },
1344
+ {
1345
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1346
+ "params": {
1347
+ "order": 0,
1348
+ "sigma": 2
1349
+ }
1350
+ }
1351
+ ]
1352
+ },
1353
+ {
1354
+ "_or_": [
1355
+ null,
1356
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
1357
+ "nirs4all.operators.transforms.signal.Detrend"
1358
+ ]
1359
+ },
1360
+ {
1361
+ "_or_": [
1362
+ null,
1363
+ "nirs4all.operators.transforms.orthogonalization.OSC",
1364
+ {
1365
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1366
+ "params": {
1367
+ "n_components": 2
1368
+ }
1369
+ },
1370
+ {
1371
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1372
+ "params": {
1373
+ "n_components": 3
1374
+ }
1375
+ }
1376
+ ]
1377
+ }
1378
+ ],
1379
+ "count": 40
1380
+ },
1381
+ {
1382
+ "class": "nirs4all.operators.splitters.splitters.SPXYFold",
1383
+ "params": {
1384
+ "n_splits": 3,
1385
+ "random_state": 42
1386
+ }
1387
+ },
1388
+ "sklearn.preprocessing._data.StandardScaler",
1389
+ {
1390
+ "model": {
1391
+ "class": "lightgbm.LGBMRegressor",
1392
+ "params": {
1393
+ "random_state": 42,
1394
+ "verbose": -1
1395
+ }
1396
+ },
1397
+ "name": "LightGBM-Sweep",
1398
+ "finetune_params": {
1399
+ "n_trials": 12,
1400
+ "sampler": "grid",
1401
+ "approach": "single",
1402
+ "eval_mode": "best",
1403
+ "verbose": 0,
1404
+ "model_params": {
1405
+ "n_estimators": [
1406
+ 150,
1407
+ 300,
1408
+ 600
1409
+ ],
1410
+ "max_depth": [
1411
+ 6,
1412
+ 10
1413
+ ],
1414
+ "learning_rate": [
1415
+ 0.03,
1416
+ 0.05
1417
+ ]
1418
+ }
1419
+ }
1420
+ },
1421
+ {
1422
+ "model": {
1423
+ "class": "xgboost.XGBRegressor",
1424
+ "params": {
1425
+ "random_state": 42,
1426
+ "verbosity": 0
1427
+ }
1428
+ },
1429
+ "name": "XGBoost-Sweep",
1430
+ "finetune_params": {
1431
+ "n_trials": 18,
1432
+ "sampler": "grid",
1433
+ "approach": "single",
1434
+ "eval_mode": "best",
1435
+ "verbose": 0,
1436
+ "model_params": {
1437
+ "n_estimators": [
1438
+ 150,
1439
+ 300,
1440
+ 600
1441
+ ],
1442
+ "max_depth": [
1443
+ 4,
1444
+ 6,
1445
+ 8
1446
+ ],
1447
+ "learning_rate": [
1448
+ 0.03,
1449
+ 0.05
1450
+ ]
1451
+ }
1452
+ }
1453
+ },
1454
+ {
1455
+ "model": {
1456
+ "function": "nirs4all.operators.models.pytorch.nicon.customizable_nicon",
1457
+ "framework": "pytorch"
1458
+ },
1459
+ "name": "NICON-CNN",
1460
+ "train_params": {
1461
+ "epochs": 250,
1462
+ "patience": 60,
1463
+ "cyclic_lr": true,
1464
+ "cyclic_lr_mode": "triangular2",
1465
+ "base_lr": 0.0005,
1466
+ "max_lr": 0.01,
1467
+ "step_size": 120
1468
+ }
1469
+ },
1470
+ {
1471
+ "model": {
1472
+ "class": "tabpfn.TabPFNRegressor",
1473
+ "params": {
1474
+ "ignore_pretraining_limits": true
1475
+ }
1476
+ },
1477
+ "name": "TabPFN"
1478
+ }
1479
+ ]
1480
+ },
1481
+ "classification": {
1482
+ "format": "yaml",
1483
+ "pipeline": [
1484
+ {
1485
+ "_cartesian_": [
1486
+ {
1487
+ "_or_": [
1488
+ null,
1489
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
1490
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
1491
+ {
1492
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
1493
+ "params": {
1494
+ "degree": 1
1495
+ }
1496
+ },
1497
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
1498
+ ]
1499
+ },
1500
+ {
1501
+ "_or_": [
1502
+ null,
1503
+ {
1504
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1505
+ "params": {
1506
+ "polyorder": 2,
1507
+ "deriv": 1
1508
+ }
1509
+ },
1510
+ {
1511
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1512
+ "params": {
1513
+ "window_length": 15,
1514
+ "polyorder": 2,
1515
+ "deriv": 1
1516
+ }
1517
+ },
1518
+ {
1519
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1520
+ "params": {
1521
+ "window_length": 21,
1522
+ "polyorder": 2,
1523
+ "deriv": 1
1524
+ }
1525
+ },
1526
+ {
1527
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1528
+ "params": {
1529
+ "window_length": 31,
1530
+ "polyorder": 2,
1531
+ "deriv": 1
1532
+ }
1533
+ },
1534
+ {
1535
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1536
+ "params": {
1537
+ "window_length": 15,
1538
+ "deriv": 2
1539
+ }
1540
+ },
1541
+ {
1542
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1543
+ "params": {
1544
+ "window_length": 21,
1545
+ "deriv": 2
1546
+ }
1547
+ },
1548
+ {
1549
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1550
+ "params": {
1551
+ "window_length": 31,
1552
+ "deriv": 2
1553
+ }
1554
+ },
1555
+ {
1556
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1557
+ "params": {
1558
+ "order": 0
1559
+ }
1560
+ },
1561
+ {
1562
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1563
+ "params": {
1564
+ "order": 0,
1565
+ "sigma": 2
1566
+ }
1567
+ }
1568
+ ]
1569
+ },
1570
+ {
1571
+ "_or_": [
1572
+ null,
1573
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
1574
+ "nirs4all.operators.transforms.signal.Detrend"
1575
+ ]
1576
+ },
1577
+ {
1578
+ "_or_": [
1579
+ null,
1580
+ "nirs4all.operators.transforms.orthogonalization.OSC",
1581
+ {
1582
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1583
+ "params": {
1584
+ "n_components": 2
1585
+ }
1586
+ },
1587
+ {
1588
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1589
+ "params": {
1590
+ "n_components": 3
1591
+ }
1592
+ }
1593
+ ]
1594
+ }
1595
+ ],
1596
+ "count": 40
1597
+ },
1598
+ {
1599
+ "class": "sklearn.model_selection._split.StratifiedKFold",
1600
+ "params": {
1601
+ "n_splits": 3,
1602
+ "shuffle": true,
1603
+ "random_state": 42
1604
+ }
1605
+ },
1606
+ "sklearn.preprocessing._data.StandardScaler",
1607
+ {
1608
+ "model": {
1609
+ "class": "lightgbm.LGBMClassifier",
1610
+ "params": {
1611
+ "random_state": 42,
1612
+ "verbose": -1
1613
+ }
1614
+ },
1615
+ "name": "LightGBM-Sweep",
1616
+ "finetune_params": {
1617
+ "n_trials": 12,
1618
+ "sampler": "grid",
1619
+ "approach": "single",
1620
+ "eval_mode": "best",
1621
+ "verbose": 0,
1622
+ "model_params": {
1623
+ "n_estimators": [
1624
+ 150,
1625
+ 300,
1626
+ 600
1627
+ ],
1628
+ "max_depth": [
1629
+ 6,
1630
+ 10
1631
+ ],
1632
+ "learning_rate": [
1633
+ 0.03,
1634
+ 0.05
1635
+ ]
1636
+ }
1637
+ }
1638
+ },
1639
+ {
1640
+ "model": {
1641
+ "class": "xgboost.XGBClassifier",
1642
+ "params": {
1643
+ "random_state": 42,
1644
+ "verbosity": 0
1645
+ }
1646
+ },
1647
+ "name": "XGBoost-Sweep",
1648
+ "finetune_params": {
1649
+ "n_trials": 18,
1650
+ "sampler": "grid",
1651
+ "approach": "single",
1652
+ "eval_mode": "best",
1653
+ "verbose": 0,
1654
+ "model_params": {
1655
+ "n_estimators": [
1656
+ 150,
1657
+ 300,
1658
+ 600
1659
+ ],
1660
+ "max_depth": [
1661
+ 4,
1662
+ 6,
1663
+ 8
1664
+ ],
1665
+ "learning_rate": [
1666
+ 0.03,
1667
+ 0.05
1668
+ ]
1669
+ }
1670
+ }
1671
+ },
1672
+ {
1673
+ "model": {
1674
+ "function": "nirs4all.operators.models.pytorch.nicon.customizable_nicon_classification",
1675
+ "framework": "pytorch"
1676
+ },
1677
+ "name": "NICON-CNN",
1678
+ "train_params": {
1679
+ "epochs": 250,
1680
+ "patience": 60,
1681
+ "cyclic_lr": true,
1682
+ "cyclic_lr_mode": "triangular2",
1683
+ "base_lr": 0.0005,
1684
+ "max_lr": 0.01,
1685
+ "step_size": 120
1686
+ }
1687
+ },
1688
+ {
1689
+ "model": {
1690
+ "class": "tabpfn.TabPFNClassifier",
1691
+ "params": {
1692
+ "ignore_pretraining_limits": true
1693
+ }
1694
+ },
1695
+ "name": "TabPFN"
1696
+ }
1697
+ ]
1698
+ }
1699
+ },
1700
+ "available_variants": [
1701
+ "regression",
1702
+ "classification"
1703
+ ],
1704
+ "steps_count": 7
1705
+ },
1706
+ {
1707
+ "id": "ultra_pls",
1708
+ "name": "Ultra PLS",
1709
+ "description": "Large cartesian chemometric exploration with PLS, OPLS, IKPLS-style linear models, and tuned Ridge.",
1710
+ "complexity": 7,
1711
+ "default_variant": "regression",
1712
+ "variants": {
1713
+ "regression": {
1714
+ "format": "yaml",
1715
+ "pipeline": [
1716
+ {
1717
+ "_cartesian_": [
1718
+ {
1719
+ "_or_": [
1720
+ null,
1721
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
1722
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
1723
+ {
1724
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
1725
+ "params": {
1726
+ "degree": 1
1727
+ }
1728
+ },
1729
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
1730
+ ]
1731
+ },
1732
+ {
1733
+ "_or_": [
1734
+ null,
1735
+ {
1736
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1737
+ "params": {
1738
+ "polyorder": 2,
1739
+ "deriv": 1
1740
+ }
1741
+ },
1742
+ {
1743
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1744
+ "params": {
1745
+ "window_length": 15,
1746
+ "polyorder": 2,
1747
+ "deriv": 1
1748
+ }
1749
+ },
1750
+ {
1751
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1752
+ "params": {
1753
+ "window_length": 21,
1754
+ "polyorder": 2,
1755
+ "deriv": 1
1756
+ }
1757
+ },
1758
+ {
1759
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1760
+ "params": {
1761
+ "window_length": 31,
1762
+ "polyorder": 2,
1763
+ "deriv": 1
1764
+ }
1765
+ },
1766
+ {
1767
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1768
+ "params": {
1769
+ "deriv": 2
1770
+ }
1771
+ },
1772
+ {
1773
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1774
+ "params": {
1775
+ "window_length": 15,
1776
+ "deriv": 2
1777
+ }
1778
+ },
1779
+ {
1780
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1781
+ "params": {
1782
+ "window_length": 21,
1783
+ "deriv": 2
1784
+ }
1785
+ },
1786
+ {
1787
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1788
+ "params": {
1789
+ "window_length": 31,
1790
+ "deriv": 2
1791
+ }
1792
+ },
1793
+ {
1794
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1795
+ "params": {
1796
+ "order": 0
1797
+ }
1798
+ },
1799
+ {
1800
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
1801
+ "params": {
1802
+ "order": 0,
1803
+ "sigma": 2
1804
+ }
1805
+ },
1806
+ {
1807
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
1808
+ "params": {
1809
+ "level": 3
1810
+ }
1811
+ },
1812
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
1813
+ ]
1814
+ },
1815
+ {
1816
+ "_or_": [
1817
+ null,
1818
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
1819
+ {
1820
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
1821
+ "params": {
1822
+ "lam": 100000.0
1823
+ }
1824
+ },
1825
+ "nirs4all.operators.transforms.signal.Detrend"
1826
+ ]
1827
+ },
1828
+ {
1829
+ "_or_": [
1830
+ null,
1831
+ "nirs4all.operators.transforms.orthogonalization.OSC",
1832
+ {
1833
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1834
+ "params": {
1835
+ "n_components": 2
1836
+ }
1837
+ },
1838
+ {
1839
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
1840
+ "params": {
1841
+ "n_components": 3
1842
+ }
1843
+ },
1844
+ {
1845
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
1846
+ "params": {
1847
+ "n_components": 0.25
1848
+ }
1849
+ }
1850
+ ]
1851
+ }
1852
+ ],
1853
+ "count": 240
1854
+ },
1855
+ {
1856
+ "class": "sklearn.model_selection._split.KFold",
1857
+ "params": {
1858
+ "n_splits": 3,
1859
+ "shuffle": true,
1860
+ "random_state": 42
1861
+ }
1862
+ },
1863
+ {
1864
+ "class": "sklearn.preprocessing._data.StandardScaler",
1865
+ "params": {
1866
+ "with_std": false
1867
+ }
1868
+ },
1869
+ {
1870
+ "model": {
1871
+ "class": "sklearn.cross_decomposition._pls.PLSRegression",
1872
+ "params": {
1873
+ "n_components": 15,
1874
+ "scale": false
1875
+ }
1876
+ },
1877
+ "name": "PLS-15"
1878
+ },
1879
+ {
1880
+ "model": {
1881
+ "class": "nirs4all.operators.models.sklearn.opls.OPLS",
1882
+ "params": {
1883
+ "n_components": 2,
1884
+ "pls_components": 15,
1885
+ "backend": "numpy"
1886
+ }
1887
+ },
1888
+ "name": "OPLS"
1889
+ },
1890
+ {
1891
+ "model": {
1892
+ "class": "nirs4all.operators.models.sklearn.ikpls.IKPLS",
1893
+ "params": {
1894
+ "n_components": 15,
1895
+ "backend": "numpy"
1896
+ }
1897
+ },
1898
+ "name": "IKPLS"
1899
+ },
1900
+ {
1901
+ "model": "sklearn.linear_model._ridge.Ridge",
1902
+ "name": "Ridge-Finetuned",
1903
+ "finetune_params": {
1904
+ "n_trials": 60,
1905
+ "sampler": "tpe",
1906
+ "model_params": {
1907
+ "alpha": [
1908
+ "log_float",
1909
+ 1e-05,
1910
+ 10000.0
1911
+ ],
1912
+ "fit_intercept": [
1913
+ "categorical",
1914
+ [
1915
+ true,
1916
+ false
1917
+ ]
1918
+ ],
1919
+ "solver": [
1920
+ "categorical",
1921
+ [
1922
+ "auto",
1923
+ "svd",
1924
+ "cholesky",
1925
+ "lsqr"
1926
+ ]
1927
+ ]
1928
+ }
1929
+ }
1930
+ }
1931
+ ]
1932
+ },
1933
+ "classification": {
1934
+ "format": "yaml",
1935
+ "pipeline": [
1936
+ {
1937
+ "_cartesian_": [
1938
+ {
1939
+ "_or_": [
1940
+ null,
1941
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
1942
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
1943
+ {
1944
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
1945
+ "params": {
1946
+ "degree": 1
1947
+ }
1948
+ },
1949
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
1950
+ ]
1951
+ },
1952
+ {
1953
+ "_or_": [
1954
+ null,
1955
+ {
1956
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1957
+ "params": {
1958
+ "polyorder": 2,
1959
+ "deriv": 1
1960
+ }
1961
+ },
1962
+ {
1963
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1964
+ "params": {
1965
+ "window_length": 15,
1966
+ "polyorder": 2,
1967
+ "deriv": 1
1968
+ }
1969
+ },
1970
+ {
1971
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1972
+ "params": {
1973
+ "window_length": 21,
1974
+ "polyorder": 2,
1975
+ "deriv": 1
1976
+ }
1977
+ },
1978
+ {
1979
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1980
+ "params": {
1981
+ "window_length": 31,
1982
+ "polyorder": 2,
1983
+ "deriv": 1
1984
+ }
1985
+ },
1986
+ {
1987
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1988
+ "params": {
1989
+ "deriv": 2
1990
+ }
1991
+ },
1992
+ {
1993
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
1994
+ "params": {
1995
+ "window_length": 15,
1996
+ "deriv": 2
1997
+ }
1998
+ },
1999
+ {
2000
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2001
+ "params": {
2002
+ "window_length": 21,
2003
+ "deriv": 2
2004
+ }
2005
+ },
2006
+ {
2007
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2008
+ "params": {
2009
+ "window_length": 31,
2010
+ "deriv": 2
2011
+ }
2012
+ },
2013
+ {
2014
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2015
+ "params": {
2016
+ "order": 0
2017
+ }
2018
+ },
2019
+ {
2020
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2021
+ "params": {
2022
+ "order": 0,
2023
+ "sigma": 2
2024
+ }
2025
+ },
2026
+ {
2027
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
2028
+ "params": {
2029
+ "level": 3
2030
+ }
2031
+ },
2032
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
2033
+ ]
2034
+ },
2035
+ {
2036
+ "_or_": [
2037
+ null,
2038
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
2039
+ {
2040
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
2041
+ "params": {
2042
+ "lam": 100000.0
2043
+ }
2044
+ },
2045
+ "nirs4all.operators.transforms.signal.Detrend"
2046
+ ]
2047
+ },
2048
+ {
2049
+ "_or_": [
2050
+ null,
2051
+ "nirs4all.operators.transforms.orthogonalization.OSC",
2052
+ {
2053
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2054
+ "params": {
2055
+ "n_components": 2
2056
+ }
2057
+ },
2058
+ {
2059
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2060
+ "params": {
2061
+ "n_components": 3
2062
+ }
2063
+ },
2064
+ {
2065
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
2066
+ "params": {
2067
+ "n_components": 0.25
2068
+ }
2069
+ }
2070
+ ]
2071
+ }
2072
+ ],
2073
+ "count": 240
2074
+ },
2075
+ {
2076
+ "class": "sklearn.model_selection._split.StratifiedKFold",
2077
+ "params": {
2078
+ "n_splits": 3,
2079
+ "shuffle": true,
2080
+ "random_state": 42
2081
+ }
2082
+ },
2083
+ {
2084
+ "class": "sklearn.preprocessing._data.StandardScaler",
2085
+ "params": {
2086
+ "with_std": false
2087
+ }
2088
+ },
2089
+ {
2090
+ "model": {
2091
+ "class": "nirs4all.operators.models.sklearn.plsda.PLSDA",
2092
+ "params": {
2093
+ "n_components": 15
2094
+ }
2095
+ },
2096
+ "name": "PLSDA-15"
2097
+ },
2098
+ {
2099
+ "model": {
2100
+ "class": "nirs4all.operators.models.sklearn.oplsda.OPLSDA",
2101
+ "params": {
2102
+ "n_components": 1,
2103
+ "pls_components": 15
2104
+ }
2105
+ },
2106
+ "name": "OPLSDA"
2107
+ },
2108
+ {
2109
+ "model": {
2110
+ "class": "nirs4all.operators.models.sklearn.aom_pls_classifier.AOMPLSClassifier",
2111
+ "params": {
2112
+ "n_components": 15
2113
+ }
2114
+ },
2115
+ "name": "AOM-PLS"
2116
+ },
2117
+ {
2118
+ "model": "sklearn.linear_model._ridge.RidgeClassifier",
2119
+ "name": "RidgeClassifier-Finetuned",
2120
+ "finetune_params": {
2121
+ "n_trials": 40,
2122
+ "sampler": "tpe",
2123
+ "model_params": {
2124
+ "alpha": [
2125
+ "log_float",
2126
+ 1e-05,
2127
+ 10000.0
2128
+ ],
2129
+ "fit_intercept": [
2130
+ "categorical",
2131
+ [
2132
+ true,
2133
+ false
2134
+ ]
2135
+ ]
2136
+ }
2137
+ }
2138
+ }
2139
+ ]
2140
+ }
2141
+ },
2142
+ "available_variants": [
2143
+ "regression",
2144
+ "classification"
2145
+ ],
2146
+ "steps_count": 7
2147
+ },
2148
+ {
2149
+ "id": "ultra_trees",
2150
+ "name": "Ultra Trees",
2151
+ "description": "Bigger preprocessing search with tuned LightGBM, RandomForest, XGBoost, ExtraTrees, and GradientBoosting.",
2152
+ "complexity": 8,
2153
+ "default_variant": "regression",
2154
+ "variants": {
2155
+ "regression": {
2156
+ "format": "yaml",
2157
+ "pipeline": [
2158
+ {
2159
+ "_cartesian_": [
2160
+ {
2161
+ "_or_": [
2162
+ null,
2163
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
2164
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
2165
+ {
2166
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
2167
+ "params": {
2168
+ "degree": 1
2169
+ }
2170
+ },
2171
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
2172
+ ]
2173
+ },
2174
+ {
2175
+ "_or_": [
2176
+ null,
2177
+ {
2178
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2179
+ "params": {
2180
+ "polyorder": 2,
2181
+ "deriv": 1
2182
+ }
2183
+ },
2184
+ {
2185
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2186
+ "params": {
2187
+ "window_length": 15,
2188
+ "polyorder": 2,
2189
+ "deriv": 1
2190
+ }
2191
+ },
2192
+ {
2193
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2194
+ "params": {
2195
+ "window_length": 21,
2196
+ "polyorder": 2,
2197
+ "deriv": 1
2198
+ }
2199
+ },
2200
+ {
2201
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2202
+ "params": {
2203
+ "window_length": 31,
2204
+ "polyorder": 2,
2205
+ "deriv": 1
2206
+ }
2207
+ },
2208
+ {
2209
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2210
+ "params": {
2211
+ "deriv": 2
2212
+ }
2213
+ },
2214
+ {
2215
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2216
+ "params": {
2217
+ "window_length": 15,
2218
+ "deriv": 2
2219
+ }
2220
+ },
2221
+ {
2222
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2223
+ "params": {
2224
+ "window_length": 21,
2225
+ "deriv": 2
2226
+ }
2227
+ },
2228
+ {
2229
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2230
+ "params": {
2231
+ "window_length": 31,
2232
+ "deriv": 2
2233
+ }
2234
+ },
2235
+ {
2236
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2237
+ "params": {
2238
+ "order": 0
2239
+ }
2240
+ },
2241
+ {
2242
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2243
+ "params": {
2244
+ "order": 0,
2245
+ "sigma": 2
2246
+ }
2247
+ },
2248
+ {
2249
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
2250
+ "params": {
2251
+ "level": 3
2252
+ }
2253
+ },
2254
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
2255
+ ]
2256
+ },
2257
+ {
2258
+ "_or_": [
2259
+ null,
2260
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
2261
+ {
2262
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
2263
+ "params": {
2264
+ "lam": 100000.0
2265
+ }
2266
+ },
2267
+ "nirs4all.operators.transforms.signal.Detrend"
2268
+ ]
2269
+ },
2270
+ {
2271
+ "_or_": [
2272
+ null,
2273
+ "nirs4all.operators.transforms.orthogonalization.OSC",
2274
+ {
2275
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2276
+ "params": {
2277
+ "n_components": 2
2278
+ }
2279
+ },
2280
+ {
2281
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2282
+ "params": {
2283
+ "n_components": 3
2284
+ }
2285
+ },
2286
+ {
2287
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
2288
+ "params": {
2289
+ "n_components": 0.25
2290
+ }
2291
+ }
2292
+ ]
2293
+ }
2294
+ ],
2295
+ "count": 80
2296
+ },
2297
+ {
2298
+ "class": "nirs4all.operators.splitters.splitters.SPXYFold",
2299
+ "params": {
2300
+ "n_splits": 3,
2301
+ "random_state": 42
2302
+ }
2303
+ },
2304
+ {
2305
+ "model": {
2306
+ "class": "lightgbm.LGBMRegressor",
2307
+ "params": {
2308
+ "random_state": 42,
2309
+ "verbose": -1
2310
+ }
2311
+ },
2312
+ "name": "LightGBM-Finetuned",
2313
+ "finetune_params": {
2314
+ "n_trials": 35,
2315
+ "sampler": "tpe",
2316
+ "model_params": {
2317
+ "n_estimators": [
2318
+ "int",
2319
+ 100,
2320
+ 600
2321
+ ],
2322
+ "max_depth": [
2323
+ "int",
2324
+ 3,
2325
+ 12
2326
+ ],
2327
+ "learning_rate": [
2328
+ "log_float",
2329
+ 0.01,
2330
+ 0.2
2331
+ ],
2332
+ "num_leaves": [
2333
+ "int",
2334
+ 15,
2335
+ 127
2336
+ ],
2337
+ "subsample": [
2338
+ "float",
2339
+ 0.7,
2340
+ 1.0
2341
+ ]
2342
+ }
2343
+ }
2344
+ },
2345
+ {
2346
+ "model": {
2347
+ "class": "sklearn.ensemble._forest.RandomForestRegressor",
2348
+ "params": {
2349
+ "random_state": 42
2350
+ }
2351
+ },
2352
+ "name": "RandomForest-Finetuned",
2353
+ "finetune_params": {
2354
+ "n_trials": 25,
2355
+ "sampler": "tpe",
2356
+ "model_params": {
2357
+ "n_estimators": [
2358
+ "int",
2359
+ 200,
2360
+ 800
2361
+ ],
2362
+ "max_depth": [
2363
+ "int",
2364
+ 4,
2365
+ 20
2366
+ ],
2367
+ "min_samples_leaf": [
2368
+ "int",
2369
+ 1,
2370
+ 5
2371
+ ],
2372
+ "min_samples_split": [
2373
+ "int",
2374
+ 2,
2375
+ 12
2376
+ ]
2377
+ }
2378
+ }
2379
+ },
2380
+ {
2381
+ "model": {
2382
+ "class": "xgboost.XGBRegressor",
2383
+ "params": {
2384
+ "random_state": 42,
2385
+ "verbosity": 0
2386
+ }
2387
+ },
2388
+ "name": "XGBoost-Finetuned",
2389
+ "finetune_params": {
2390
+ "n_trials": 35,
2391
+ "sampler": "tpe",
2392
+ "model_params": {
2393
+ "n_estimators": [
2394
+ "int",
2395
+ 100,
2396
+ 600
2397
+ ],
2398
+ "max_depth": [
2399
+ "int",
2400
+ 3,
2401
+ 10
2402
+ ],
2403
+ "learning_rate": [
2404
+ "log_float",
2405
+ 0.01,
2406
+ 0.2
2407
+ ],
2408
+ "subsample": [
2409
+ "float",
2410
+ 0.7,
2411
+ 1.0
2412
+ ],
2413
+ "colsample_bytree": [
2414
+ "float",
2415
+ 0.6,
2416
+ 1.0
2417
+ ]
2418
+ }
2419
+ }
2420
+ },
2421
+ {
2422
+ "model": {
2423
+ "class": "sklearn.ensemble._forest.ExtraTreesRegressor",
2424
+ "params": {
2425
+ "random_state": 42
2426
+ }
2427
+ },
2428
+ "name": "ExtraTrees-Finetuned",
2429
+ "finetune_params": {
2430
+ "n_trials": 20,
2431
+ "sampler": "tpe",
2432
+ "model_params": {
2433
+ "n_estimators": [
2434
+ "int",
2435
+ 200,
2436
+ 800
2437
+ ],
2438
+ "max_depth": [
2439
+ "int",
2440
+ 4,
2441
+ 20
2442
+ ],
2443
+ "min_samples_leaf": [
2444
+ "int",
2445
+ 1,
2446
+ 5
2447
+ ]
2448
+ }
2449
+ }
2450
+ },
2451
+ {
2452
+ "model": {
2453
+ "class": "sklearn.ensemble._gb.GradientBoostingRegressor",
2454
+ "params": {
2455
+ "random_state": 42
2456
+ }
2457
+ },
2458
+ "name": "GradientBoosting-Finetuned",
2459
+ "finetune_params": {
2460
+ "n_trials": 20,
2461
+ "sampler": "tpe",
2462
+ "model_params": {
2463
+ "n_estimators": [
2464
+ "int",
2465
+ 100,
2466
+ 500
2467
+ ],
2468
+ "learning_rate": [
2469
+ "log_float",
2470
+ 0.01,
2471
+ 0.2
2472
+ ],
2473
+ "max_depth": [
2474
+ "int",
2475
+ 2,
2476
+ 6
2477
+ ],
2478
+ "subsample": [
2479
+ "float",
2480
+ 0.7,
2481
+ 1.0
2482
+ ]
2483
+ }
2484
+ }
2485
+ }
2486
+ ]
2487
+ },
2488
+ "classification": {
2489
+ "format": "yaml",
2490
+ "pipeline": [
2491
+ {
2492
+ "_cartesian_": [
2493
+ {
2494
+ "_or_": [
2495
+ null,
2496
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
2497
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
2498
+ {
2499
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
2500
+ "params": {
2501
+ "degree": 1
2502
+ }
2503
+ },
2504
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
2505
+ ]
2506
+ },
2507
+ {
2508
+ "_or_": [
2509
+ null,
2510
+ {
2511
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2512
+ "params": {
2513
+ "polyorder": 2,
2514
+ "deriv": 1
2515
+ }
2516
+ },
2517
+ {
2518
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2519
+ "params": {
2520
+ "window_length": 15,
2521
+ "polyorder": 2,
2522
+ "deriv": 1
2523
+ }
2524
+ },
2525
+ {
2526
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2527
+ "params": {
2528
+ "window_length": 21,
2529
+ "polyorder": 2,
2530
+ "deriv": 1
2531
+ }
2532
+ },
2533
+ {
2534
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2535
+ "params": {
2536
+ "window_length": 31,
2537
+ "polyorder": 2,
2538
+ "deriv": 1
2539
+ }
2540
+ },
2541
+ {
2542
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2543
+ "params": {
2544
+ "deriv": 2
2545
+ }
2546
+ },
2547
+ {
2548
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2549
+ "params": {
2550
+ "window_length": 15,
2551
+ "deriv": 2
2552
+ }
2553
+ },
2554
+ {
2555
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2556
+ "params": {
2557
+ "window_length": 21,
2558
+ "deriv": 2
2559
+ }
2560
+ },
2561
+ {
2562
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2563
+ "params": {
2564
+ "window_length": 31,
2565
+ "deriv": 2
2566
+ }
2567
+ },
2568
+ {
2569
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2570
+ "params": {
2571
+ "order": 0
2572
+ }
2573
+ },
2574
+ {
2575
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2576
+ "params": {
2577
+ "order": 0,
2578
+ "sigma": 2
2579
+ }
2580
+ },
2581
+ {
2582
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
2583
+ "params": {
2584
+ "level": 3
2585
+ }
2586
+ },
2587
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
2588
+ ]
2589
+ },
2590
+ {
2591
+ "_or_": [
2592
+ null,
2593
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
2594
+ {
2595
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
2596
+ "params": {
2597
+ "lam": 100000.0
2598
+ }
2599
+ },
2600
+ "nirs4all.operators.transforms.signal.Detrend"
2601
+ ]
2602
+ },
2603
+ {
2604
+ "_or_": [
2605
+ null,
2606
+ "nirs4all.operators.transforms.orthogonalization.OSC",
2607
+ {
2608
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2609
+ "params": {
2610
+ "n_components": 2
2611
+ }
2612
+ },
2613
+ {
2614
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2615
+ "params": {
2616
+ "n_components": 3
2617
+ }
2618
+ },
2619
+ {
2620
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
2621
+ "params": {
2622
+ "n_components": 0.25
2623
+ }
2624
+ }
2625
+ ]
2626
+ }
2627
+ ],
2628
+ "count": 80
2629
+ },
2630
+ {
2631
+ "class": "sklearn.model_selection._split.StratifiedKFold",
2632
+ "params": {
2633
+ "n_splits": 3,
2634
+ "shuffle": true,
2635
+ "random_state": 42
2636
+ }
2637
+ },
2638
+ {
2639
+ "model": {
2640
+ "class": "lightgbm.LGBMClassifier",
2641
+ "params": {
2642
+ "random_state": 42,
2643
+ "verbose": -1
2644
+ }
2645
+ },
2646
+ "name": "LightGBM-Finetuned",
2647
+ "finetune_params": {
2648
+ "n_trials": 35,
2649
+ "sampler": "tpe",
2650
+ "model_params": {
2651
+ "n_estimators": [
2652
+ "int",
2653
+ 100,
2654
+ 600
2655
+ ],
2656
+ "max_depth": [
2657
+ "int",
2658
+ 3,
2659
+ 12
2660
+ ],
2661
+ "learning_rate": [
2662
+ "log_float",
2663
+ 0.01,
2664
+ 0.2
2665
+ ],
2666
+ "num_leaves": [
2667
+ "int",
2668
+ 15,
2669
+ 127
2670
+ ],
2671
+ "subsample": [
2672
+ "float",
2673
+ 0.7,
2674
+ 1.0
2675
+ ]
2676
+ }
2677
+ }
2678
+ },
2679
+ {
2680
+ "model": {
2681
+ "class": "sklearn.ensemble._forest.RandomForestClassifier",
2682
+ "params": {
2683
+ "random_state": 42
2684
+ }
2685
+ },
2686
+ "name": "RandomForest-Finetuned",
2687
+ "finetune_params": {
2688
+ "n_trials": 25,
2689
+ "sampler": "tpe",
2690
+ "model_params": {
2691
+ "n_estimators": [
2692
+ "int",
2693
+ 200,
2694
+ 800
2695
+ ],
2696
+ "max_depth": [
2697
+ "int",
2698
+ 4,
2699
+ 20
2700
+ ],
2701
+ "min_samples_leaf": [
2702
+ "int",
2703
+ 1,
2704
+ 5
2705
+ ],
2706
+ "min_samples_split": [
2707
+ "int",
2708
+ 2,
2709
+ 12
2710
+ ]
2711
+ }
2712
+ }
2713
+ },
2714
+ {
2715
+ "model": {
2716
+ "class": "xgboost.XGBClassifier",
2717
+ "params": {
2718
+ "random_state": 42,
2719
+ "verbosity": 0
2720
+ }
2721
+ },
2722
+ "name": "XGBoost-Finetuned",
2723
+ "finetune_params": {
2724
+ "n_trials": 35,
2725
+ "sampler": "tpe",
2726
+ "model_params": {
2727
+ "n_estimators": [
2728
+ "int",
2729
+ 100,
2730
+ 600
2731
+ ],
2732
+ "max_depth": [
2733
+ "int",
2734
+ 3,
2735
+ 10
2736
+ ],
2737
+ "learning_rate": [
2738
+ "log_float",
2739
+ 0.01,
2740
+ 0.2
2741
+ ],
2742
+ "subsample": [
2743
+ "float",
2744
+ 0.7,
2745
+ 1.0
2746
+ ],
2747
+ "colsample_bytree": [
2748
+ "float",
2749
+ 0.6,
2750
+ 1.0
2751
+ ]
2752
+ }
2753
+ }
2754
+ },
2755
+ {
2756
+ "model": {
2757
+ "class": "sklearn.ensemble._forest.ExtraTreesClassifier",
2758
+ "params": {
2759
+ "random_state": 42
2760
+ }
2761
+ },
2762
+ "name": "ExtraTrees-Finetuned",
2763
+ "finetune_params": {
2764
+ "n_trials": 20,
2765
+ "sampler": "tpe",
2766
+ "model_params": {
2767
+ "n_estimators": [
2768
+ "int",
2769
+ 200,
2770
+ 800
2771
+ ],
2772
+ "max_depth": [
2773
+ "int",
2774
+ 4,
2775
+ 20
2776
+ ],
2777
+ "min_samples_leaf": [
2778
+ "int",
2779
+ 1,
2780
+ 5
2781
+ ]
2782
+ }
2783
+ }
2784
+ },
2785
+ {
2786
+ "model": {
2787
+ "class": "sklearn.ensemble._gb.GradientBoostingClassifier",
2788
+ "params": {
2789
+ "random_state": 42
2790
+ }
2791
+ },
2792
+ "name": "GradientBoosting-Finetuned",
2793
+ "finetune_params": {
2794
+ "n_trials": 20,
2795
+ "sampler": "tpe",
2796
+ "model_params": {
2797
+ "n_estimators": [
2798
+ "int",
2799
+ 100,
2800
+ 500
2801
+ ],
2802
+ "learning_rate": [
2803
+ "log_float",
2804
+ 0.01,
2805
+ 0.2
2806
+ ],
2807
+ "max_depth": [
2808
+ "int",
2809
+ 2,
2810
+ 6
2811
+ ],
2812
+ "subsample": [
2813
+ "float",
2814
+ 0.7,
2815
+ 1.0
2816
+ ]
2817
+ }
2818
+ }
2819
+ }
2820
+ ]
2821
+ }
2822
+ },
2823
+ "available_variants": [
2824
+ "regression",
2825
+ "classification"
2826
+ ],
2827
+ "steps_count": 7
2828
+ },
2829
+ {
2830
+ "id": "deep_nonlinear_exploration",
2831
+ "name": "Deep Non-Linear Exploration",
2832
+ "description": "Large cartesian preprocessing with tuned gradient-boosted trees, tuned NICON, and TabPFN.",
2833
+ "complexity": 9,
2834
+ "default_variant": "regression",
2835
+ "variants": {
2836
+ "regression": {
2837
+ "format": "yaml",
2838
+ "pipeline": [
2839
+ {
2840
+ "_cartesian_": [
2841
+ {
2842
+ "_or_": [
2843
+ null,
2844
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
2845
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
2846
+ {
2847
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
2848
+ "params": {
2849
+ "degree": 1
2850
+ }
2851
+ },
2852
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
2853
+ ]
2854
+ },
2855
+ {
2856
+ "_or_": [
2857
+ null,
2858
+ {
2859
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2860
+ "params": {
2861
+ "polyorder": 2,
2862
+ "deriv": 1
2863
+ }
2864
+ },
2865
+ {
2866
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2867
+ "params": {
2868
+ "window_length": 15,
2869
+ "polyorder": 2,
2870
+ "deriv": 1
2871
+ }
2872
+ },
2873
+ {
2874
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2875
+ "params": {
2876
+ "window_length": 21,
2877
+ "polyorder": 2,
2878
+ "deriv": 1
2879
+ }
2880
+ },
2881
+ {
2882
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2883
+ "params": {
2884
+ "window_length": 31,
2885
+ "polyorder": 2,
2886
+ "deriv": 1
2887
+ }
2888
+ },
2889
+ {
2890
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2891
+ "params": {
2892
+ "deriv": 2
2893
+ }
2894
+ },
2895
+ {
2896
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2897
+ "params": {
2898
+ "window_length": 15,
2899
+ "deriv": 2
2900
+ }
2901
+ },
2902
+ {
2903
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2904
+ "params": {
2905
+ "window_length": 21,
2906
+ "deriv": 2
2907
+ }
2908
+ },
2909
+ {
2910
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
2911
+ "params": {
2912
+ "window_length": 31,
2913
+ "deriv": 2
2914
+ }
2915
+ },
2916
+ {
2917
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2918
+ "params": {
2919
+ "order": 0
2920
+ }
2921
+ },
2922
+ {
2923
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
2924
+ "params": {
2925
+ "order": 0,
2926
+ "sigma": 2
2927
+ }
2928
+ },
2929
+ {
2930
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
2931
+ "params": {
2932
+ "level": 3
2933
+ }
2934
+ },
2935
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
2936
+ ]
2937
+ },
2938
+ {
2939
+ "_or_": [
2940
+ null,
2941
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
2942
+ {
2943
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
2944
+ "params": {
2945
+ "lam": 100000.0
2946
+ }
2947
+ },
2948
+ "nirs4all.operators.transforms.signal.Detrend"
2949
+ ]
2950
+ },
2951
+ {
2952
+ "_or_": [
2953
+ null,
2954
+ "nirs4all.operators.transforms.orthogonalization.OSC",
2955
+ {
2956
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2957
+ "params": {
2958
+ "n_components": 2
2959
+ }
2960
+ },
2961
+ {
2962
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
2963
+ "params": {
2964
+ "n_components": 3
2965
+ }
2966
+ },
2967
+ {
2968
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
2969
+ "params": {
2970
+ "n_components": 0.25
2971
+ }
2972
+ }
2973
+ ]
2974
+ }
2975
+ ],
2976
+ "count": 80
2977
+ },
2978
+ {
2979
+ "class": "nirs4all.operators.splitters.splitters.SPXYFold",
2980
+ "params": {
2981
+ "n_splits": 3,
2982
+ "random_state": 42
2983
+ }
2984
+ },
2985
+ "sklearn.preprocessing._data.StandardScaler",
2986
+ {
2987
+ "model": {
2988
+ "class": "lightgbm.LGBMRegressor",
2989
+ "params": {
2990
+ "random_state": 42,
2991
+ "verbose": -1
2992
+ }
2993
+ },
2994
+ "name": "LightGBM-Finetuned",
2995
+ "finetune_params": {
2996
+ "n_trials": 25,
2997
+ "sampler": "tpe",
2998
+ "model_params": {
2999
+ "n_estimators": [
3000
+ "int",
3001
+ 100,
3002
+ 600
3003
+ ],
3004
+ "max_depth": [
3005
+ "int",
3006
+ 3,
3007
+ 12
3008
+ ],
3009
+ "learning_rate": [
3010
+ "log_float",
3011
+ 0.01,
3012
+ 0.2
3013
+ ],
3014
+ "num_leaves": [
3015
+ "int",
3016
+ 15,
3017
+ 127
3018
+ ],
3019
+ "subsample": [
3020
+ "float",
3021
+ 0.7,
3022
+ 1.0
3023
+ ]
3024
+ }
3025
+ }
3026
+ },
3027
+ {
3028
+ "model": {
3029
+ "class": "xgboost.XGBRegressor",
3030
+ "params": {
3031
+ "random_state": 42,
3032
+ "verbosity": 0
3033
+ }
3034
+ },
3035
+ "name": "XGBoost-Finetuned",
3036
+ "finetune_params": {
3037
+ "n_trials": 25,
3038
+ "sampler": "tpe",
3039
+ "model_params": {
3040
+ "n_estimators": [
3041
+ "int",
3042
+ 100,
3043
+ 600
3044
+ ],
3045
+ "max_depth": [
3046
+ "int",
3047
+ 3,
3048
+ 10
3049
+ ],
3050
+ "learning_rate": [
3051
+ "log_float",
3052
+ 0.01,
3053
+ 0.2
3054
+ ],
3055
+ "subsample": [
3056
+ "float",
3057
+ 0.7,
3058
+ 1.0
3059
+ ],
3060
+ "colsample_bytree": [
3061
+ "float",
3062
+ 0.6,
3063
+ 1.0
3064
+ ]
3065
+ }
3066
+ }
3067
+ },
3068
+ {
3069
+ "model": {
3070
+ "function": "nirs4all.operators.models.pytorch.nicon.customizable_nicon",
3071
+ "framework": "pytorch"
3072
+ },
3073
+ "name": "NICON-CNN",
3074
+ "finetune_params": {
3075
+ "n_trials": 25,
3076
+ "sampler": "tpe",
3077
+ "pruner": "hyperband",
3078
+ "approach": "grouped",
3079
+ "model_params": {
3080
+ "spatial_dropout": [
3081
+ "float",
3082
+ 0.01,
3083
+ 0.5
3084
+ ],
3085
+ "filters1": [
3086
+ "categorical",
3087
+ [
3088
+ 4,
3089
+ 8,
3090
+ 16,
3091
+ 32
3092
+ ]
3093
+ ],
3094
+ "dropout_rate": [
3095
+ "float",
3096
+ 0.01,
3097
+ 0.5
3098
+ ],
3099
+ "filters2": [
3100
+ "categorical",
3101
+ [
3102
+ 32,
3103
+ 64,
3104
+ 128,
3105
+ 256
3106
+ ]
3107
+ ],
3108
+ "filters3": [
3109
+ "categorical",
3110
+ [
3111
+ 8,
3112
+ 16,
3113
+ 32,
3114
+ 64
3115
+ ]
3116
+ ],
3117
+ "dense_units": [
3118
+ "categorical",
3119
+ [
3120
+ 8,
3121
+ 16,
3122
+ 32,
3123
+ 64
3124
+ ]
3125
+ ]
3126
+ },
3127
+ "train_params": {
3128
+ "epochs": 60,
3129
+ "patience": 25,
3130
+ "learning_rate": 0.005,
3131
+ "verbose": 0
3132
+ }
3133
+ },
3134
+ "train_params": {
3135
+ "epochs": 400,
3136
+ "patience": 100,
3137
+ "cyclic_lr": true,
3138
+ "cyclic_lr_mode": "triangular2",
3139
+ "base_lr": 0.0005,
3140
+ "max_lr": 0.01,
3141
+ "step_size": 120
3142
+ }
3143
+ },
3144
+ {
3145
+ "model": {
3146
+ "class": "tabpfn.TabPFNRegressor",
3147
+ "params": {
3148
+ "ignore_pretraining_limits": true
3149
+ }
3150
+ },
3151
+ "name": "TabPFN"
3152
+ }
3153
+ ]
3154
+ },
3155
+ "classification": {
3156
+ "format": "yaml",
3157
+ "pipeline": [
3158
+ {
3159
+ "_cartesian_": [
3160
+ {
3161
+ "_or_": [
3162
+ null,
3163
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
3164
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
3165
+ {
3166
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
3167
+ "params": {
3168
+ "degree": 1
3169
+ }
3170
+ },
3171
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
3172
+ ]
3173
+ },
3174
+ {
3175
+ "_or_": [
3176
+ null,
3177
+ {
3178
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3179
+ "params": {
3180
+ "polyorder": 2,
3181
+ "deriv": 1
3182
+ }
3183
+ },
3184
+ {
3185
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3186
+ "params": {
3187
+ "window_length": 15,
3188
+ "polyorder": 2,
3189
+ "deriv": 1
3190
+ }
3191
+ },
3192
+ {
3193
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3194
+ "params": {
3195
+ "window_length": 21,
3196
+ "polyorder": 2,
3197
+ "deriv": 1
3198
+ }
3199
+ },
3200
+ {
3201
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3202
+ "params": {
3203
+ "window_length": 31,
3204
+ "polyorder": 2,
3205
+ "deriv": 1
3206
+ }
3207
+ },
3208
+ {
3209
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3210
+ "params": {
3211
+ "deriv": 2
3212
+ }
3213
+ },
3214
+ {
3215
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3216
+ "params": {
3217
+ "window_length": 15,
3218
+ "deriv": 2
3219
+ }
3220
+ },
3221
+ {
3222
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3223
+ "params": {
3224
+ "window_length": 21,
3225
+ "deriv": 2
3226
+ }
3227
+ },
3228
+ {
3229
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3230
+ "params": {
3231
+ "window_length": 31,
3232
+ "deriv": 2
3233
+ }
3234
+ },
3235
+ {
3236
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3237
+ "params": {
3238
+ "order": 0
3239
+ }
3240
+ },
3241
+ {
3242
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3243
+ "params": {
3244
+ "order": 0,
3245
+ "sigma": 2
3246
+ }
3247
+ },
3248
+ {
3249
+ "class": "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise",
3250
+ "params": {
3251
+ "level": 3
3252
+ }
3253
+ },
3254
+ "nirs4all.operators.transforms.wavelet_denoise.WaveletDenoise"
3255
+ ]
3256
+ },
3257
+ {
3258
+ "_or_": [
3259
+ null,
3260
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
3261
+ {
3262
+ "class": "nirs4all.operators.transforms.nirs.ASLSBaseline",
3263
+ "params": {
3264
+ "lam": 100000.0
3265
+ }
3266
+ },
3267
+ "nirs4all.operators.transforms.signal.Detrend"
3268
+ ]
3269
+ },
3270
+ {
3271
+ "_or_": [
3272
+ null,
3273
+ "nirs4all.operators.transforms.orthogonalization.OSC",
3274
+ {
3275
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3276
+ "params": {
3277
+ "n_components": 2
3278
+ }
3279
+ },
3280
+ {
3281
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3282
+ "params": {
3283
+ "n_components": 3
3284
+ }
3285
+ },
3286
+ {
3287
+ "class": "nirs4all.operators.transforms.feature_selection.FlexiblePCA",
3288
+ "params": {
3289
+ "n_components": 0.25
3290
+ }
3291
+ }
3292
+ ]
3293
+ }
3294
+ ],
3295
+ "count": 80
3296
+ },
3297
+ {
3298
+ "class": "sklearn.model_selection._split.StratifiedKFold",
3299
+ "params": {
3300
+ "n_splits": 3,
3301
+ "shuffle": true,
3302
+ "random_state": 42
3303
+ }
3304
+ },
3305
+ "sklearn.preprocessing._data.StandardScaler",
3306
+ {
3307
+ "model": {
3308
+ "class": "lightgbm.LGBMClassifier",
3309
+ "params": {
3310
+ "random_state": 42,
3311
+ "verbose": -1
3312
+ }
3313
+ },
3314
+ "name": "LightGBM-Finetuned",
3315
+ "finetune_params": {
3316
+ "n_trials": 25,
3317
+ "sampler": "tpe",
3318
+ "model_params": {
3319
+ "n_estimators": [
3320
+ "int",
3321
+ 100,
3322
+ 600
3323
+ ],
3324
+ "max_depth": [
3325
+ "int",
3326
+ 3,
3327
+ 12
3328
+ ],
3329
+ "learning_rate": [
3330
+ "log_float",
3331
+ 0.01,
3332
+ 0.2
3333
+ ],
3334
+ "num_leaves": [
3335
+ "int",
3336
+ 15,
3337
+ 127
3338
+ ],
3339
+ "subsample": [
3340
+ "float",
3341
+ 0.7,
3342
+ 1.0
3343
+ ]
3344
+ }
3345
+ }
3346
+ },
3347
+ {
3348
+ "model": {
3349
+ "class": "xgboost.XGBClassifier",
3350
+ "params": {
3351
+ "random_state": 42,
3352
+ "verbosity": 0
3353
+ }
3354
+ },
3355
+ "name": "XGBoost-Finetuned",
3356
+ "finetune_params": {
3357
+ "n_trials": 25,
3358
+ "sampler": "tpe",
3359
+ "model_params": {
3360
+ "n_estimators": [
3361
+ "int",
3362
+ 100,
3363
+ 600
3364
+ ],
3365
+ "max_depth": [
3366
+ "int",
3367
+ 3,
3368
+ 10
3369
+ ],
3370
+ "learning_rate": [
3371
+ "log_float",
3372
+ 0.01,
3373
+ 0.2
3374
+ ],
3375
+ "subsample": [
3376
+ "float",
3377
+ 0.7,
3378
+ 1.0
3379
+ ],
3380
+ "colsample_bytree": [
3381
+ "float",
3382
+ 0.6,
3383
+ 1.0
3384
+ ]
3385
+ }
3386
+ }
3387
+ },
3388
+ {
3389
+ "model": {
3390
+ "function": "nirs4all.operators.models.pytorch.nicon.customizable_nicon_classification",
3391
+ "framework": "pytorch"
3392
+ },
3393
+ "name": "NICON-CNN",
3394
+ "finetune_params": {
3395
+ "n_trials": 25,
3396
+ "sampler": "tpe",
3397
+ "pruner": "hyperband",
3398
+ "approach": "grouped",
3399
+ "model_params": {
3400
+ "spatial_dropout": [
3401
+ "float",
3402
+ 0.01,
3403
+ 0.5
3404
+ ],
3405
+ "filters1": [
3406
+ "categorical",
3407
+ [
3408
+ 4,
3409
+ 8,
3410
+ 16,
3411
+ 32
3412
+ ]
3413
+ ],
3414
+ "dropout_rate": [
3415
+ "float",
3416
+ 0.01,
3417
+ 0.5
3418
+ ],
3419
+ "filters2": [
3420
+ "categorical",
3421
+ [
3422
+ 32,
3423
+ 64,
3424
+ 128,
3425
+ 256
3426
+ ]
3427
+ ],
3428
+ "filters3": [
3429
+ "categorical",
3430
+ [
3431
+ 8,
3432
+ 16,
3433
+ 32,
3434
+ 64
3435
+ ]
3436
+ ],
3437
+ "dense_units": [
3438
+ "categorical",
3439
+ [
3440
+ 8,
3441
+ 16,
3442
+ 32,
3443
+ 64
3444
+ ]
3445
+ ]
3446
+ },
3447
+ "train_params": {
3448
+ "epochs": 60,
3449
+ "patience": 25,
3450
+ "learning_rate": 0.005,
3451
+ "verbose": 0
3452
+ }
3453
+ },
3454
+ "train_params": {
3455
+ "epochs": 400,
3456
+ "patience": 100,
3457
+ "cyclic_lr": true,
3458
+ "cyclic_lr_mode": "triangular2",
3459
+ "base_lr": 0.0005,
3460
+ "max_lr": 0.01,
3461
+ "step_size": 120
3462
+ }
3463
+ },
3464
+ {
3465
+ "model": {
3466
+ "class": "tabpfn.TabPFNClassifier",
3467
+ "params": {
3468
+ "ignore_pretraining_limits": true
3469
+ }
3470
+ },
3471
+ "name": "TabPFN"
3472
+ }
3473
+ ]
3474
+ }
3475
+ },
3476
+ "available_variants": [
3477
+ "regression",
3478
+ "classification"
3479
+ ],
3480
+ "steps_count": 7
3481
+ },
3482
+ {
3483
+ "id": "ultra_slow",
3484
+ "name": "Ultra Slow",
3485
+ "description": "Heavy Cartesian preset mixing tuned PLS, tuned Ridge, tuned AOM-PLS, tuned LightGBM, and TabPFN.",
3486
+ "complexity": 10,
3487
+ "default_variant": "regression",
3488
+ "variants": {
3489
+ "regression": {
3490
+ "format": "yaml",
3491
+ "pipeline": [
3492
+ {
3493
+ "_cartesian_": [
3494
+ {
3495
+ "_or_": [
3496
+ null,
3497
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
3498
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
3499
+ {
3500
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
3501
+ "params": {
3502
+ "degree": 1
3503
+ }
3504
+ },
3505
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
3506
+ ]
3507
+ },
3508
+ {
3509
+ "_or_": [
3510
+ null,
3511
+ {
3512
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3513
+ "params": {
3514
+ "polyorder": 2,
3515
+ "deriv": 1
3516
+ }
3517
+ },
3518
+ {
3519
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3520
+ "params": {
3521
+ "window_length": 15,
3522
+ "polyorder": 2,
3523
+ "deriv": 1
3524
+ }
3525
+ },
3526
+ {
3527
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3528
+ "params": {
3529
+ "window_length": 21,
3530
+ "polyorder": 2,
3531
+ "deriv": 1
3532
+ }
3533
+ },
3534
+ {
3535
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3536
+ "params": {
3537
+ "window_length": 31,
3538
+ "polyorder": 2,
3539
+ "deriv": 1
3540
+ }
3541
+ },
3542
+ {
3543
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3544
+ "params": {
3545
+ "window_length": 15,
3546
+ "deriv": 2
3547
+ }
3548
+ },
3549
+ {
3550
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3551
+ "params": {
3552
+ "window_length": 21,
3553
+ "deriv": 2
3554
+ }
3555
+ },
3556
+ {
3557
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3558
+ "params": {
3559
+ "window_length": 31,
3560
+ "deriv": 2
3561
+ }
3562
+ },
3563
+ {
3564
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3565
+ "params": {
3566
+ "order": 0
3567
+ }
3568
+ },
3569
+ {
3570
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3571
+ "params": {
3572
+ "order": 0,
3573
+ "sigma": 2
3574
+ }
3575
+ }
3576
+ ]
3577
+ },
3578
+ {
3579
+ "_or_": [
3580
+ null,
3581
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
3582
+ "nirs4all.operators.transforms.signal.Detrend"
3583
+ ]
3584
+ },
3585
+ {
3586
+ "_or_": [
3587
+ null,
3588
+ "nirs4all.operators.transforms.orthogonalization.OSC",
3589
+ {
3590
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3591
+ "params": {
3592
+ "n_components": 2
3593
+ }
3594
+ },
3595
+ {
3596
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3597
+ "params": {
3598
+ "n_components": 3
3599
+ }
3600
+ }
3601
+ ]
3602
+ }
3603
+ ],
3604
+ "count": 90
3605
+ },
3606
+ {
3607
+ "class": "nirs4all.operators.splitters.splitters.SPXYFold",
3608
+ "params": {
3609
+ "n_splits": 3,
3610
+ "random_state": 42
3611
+ }
3612
+ },
3613
+ {
3614
+ "class": "sklearn.preprocessing._data.StandardScaler",
3615
+ "params": {
3616
+ "with_std": false
3617
+ }
3618
+ },
3619
+ {
3620
+ "model": {
3621
+ "class": "sklearn.cross_decomposition._pls.PLSRegression",
3622
+ "params": {
3623
+ "scale": false
3624
+ }
3625
+ },
3626
+ "name": "PLS-Finetuned",
3627
+ "finetune_params": {
3628
+ "n_trials": 40,
3629
+ "sampler": "binary",
3630
+ "model_params": {
3631
+ "n_components": [
3632
+ "int",
3633
+ 1,
3634
+ 30
3635
+ ]
3636
+ }
3637
+ }
3638
+ },
3639
+ {
3640
+ "model": "sklearn.linear_model._ridge.Ridge",
3641
+ "name": "Ridge-Finetuned",
3642
+ "finetune_params": {
3643
+ "n_trials": 60,
3644
+ "sampler": "tpe",
3645
+ "model_params": {
3646
+ "alpha": [
3647
+ "log_float",
3648
+ 1e-05,
3649
+ 10000.0
3650
+ ],
3651
+ "fit_intercept": [
3652
+ "categorical",
3653
+ [
3654
+ true,
3655
+ false
3656
+ ]
3657
+ ],
3658
+ "solver": [
3659
+ "categorical",
3660
+ [
3661
+ "auto",
3662
+ "svd",
3663
+ "cholesky",
3664
+ "lsqr"
3665
+ ]
3666
+ ]
3667
+ }
3668
+ }
3669
+ },
3670
+ {
3671
+ "model": {
3672
+ "class": "lightgbm.LGBMRegressor",
3673
+ "params": {
3674
+ "random_state": 42,
3675
+ "verbose": -1
3676
+ }
3677
+ },
3678
+ "name": "LightGBM-Finetuned",
3679
+ "finetune_params": {
3680
+ "n_trials": 30,
3681
+ "sampler": "tpe",
3682
+ "model_params": {
3683
+ "n_estimators": [
3684
+ "int",
3685
+ 100,
3686
+ 600
3687
+ ],
3688
+ "max_depth": [
3689
+ "int",
3690
+ 3,
3691
+ 12
3692
+ ],
3693
+ "learning_rate": [
3694
+ "log_float",
3695
+ 0.01,
3696
+ 0.2
3697
+ ],
3698
+ "num_leaves": [
3699
+ "int",
3700
+ 15,
3701
+ 127
3702
+ ],
3703
+ "subsample": [
3704
+ "float",
3705
+ 0.7,
3706
+ 1.0
3707
+ ]
3708
+ }
3709
+ }
3710
+ },
3711
+ {
3712
+ "model": {
3713
+ "class": "nirs4all.operators.models.sklearn.aom_pls.AOMPLSRegressor",
3714
+ "params": {
3715
+ "center": true,
3716
+ "scale": false
3717
+ }
3718
+ },
3719
+ "name": "AOM-PLS-Finetuned",
3720
+ "finetune_params": {
3721
+ "n_trials": 80,
3722
+ "sampler": "tpe",
3723
+ "model_params": {
3724
+ "n_components": [
3725
+ "int",
3726
+ 1,
3727
+ 27
3728
+ ],
3729
+ "n_orth": [
3730
+ "int",
3731
+ 0,
3732
+ 5
3733
+ ],
3734
+ "operator_index": [
3735
+ "int",
3736
+ 0,
3737
+ 120
3738
+ ]
3739
+ }
3740
+ }
3741
+ },
3742
+ {
3743
+ "model": {
3744
+ "class": "tabpfn.TabPFNRegressor",
3745
+ "params": {
3746
+ "ignore_pretraining_limits": true
3747
+ }
3748
+ },
3749
+ "name": "TabPFN"
3750
+ }
3751
+ ]
3752
+ },
3753
+ "classification": {
3754
+ "format": "yaml",
3755
+ "pipeline": [
3756
+ {
3757
+ "_cartesian_": [
3758
+ {
3759
+ "_or_": [
3760
+ null,
3761
+ "nirs4all.operators.transforms.scalers.StandardNormalVariate",
3762
+ "nirs4all.operators.transforms.nirs.MultiplicativeScatterCorrection",
3763
+ {
3764
+ "class": "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection",
3765
+ "params": {
3766
+ "degree": 1
3767
+ }
3768
+ },
3769
+ "nirs4all.operators.transforms.nirs.ExtendedMultiplicativeScatterCorrection"
3770
+ ]
3771
+ },
3772
+ {
3773
+ "_or_": [
3774
+ null,
3775
+ {
3776
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3777
+ "params": {
3778
+ "polyorder": 2,
3779
+ "deriv": 1
3780
+ }
3781
+ },
3782
+ {
3783
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3784
+ "params": {
3785
+ "window_length": 15,
3786
+ "polyorder": 2,
3787
+ "deriv": 1
3788
+ }
3789
+ },
3790
+ {
3791
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3792
+ "params": {
3793
+ "window_length": 21,
3794
+ "polyorder": 2,
3795
+ "deriv": 1
3796
+ }
3797
+ },
3798
+ {
3799
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3800
+ "params": {
3801
+ "window_length": 31,
3802
+ "polyorder": 2,
3803
+ "deriv": 1
3804
+ }
3805
+ },
3806
+ {
3807
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3808
+ "params": {
3809
+ "window_length": 15,
3810
+ "deriv": 2
3811
+ }
3812
+ },
3813
+ {
3814
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3815
+ "params": {
3816
+ "window_length": 21,
3817
+ "deriv": 2
3818
+ }
3819
+ },
3820
+ {
3821
+ "class": "nirs4all.operators.transforms.nirs.SavitzkyGolay",
3822
+ "params": {
3823
+ "window_length": 31,
3824
+ "deriv": 2
3825
+ }
3826
+ },
3827
+ {
3828
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3829
+ "params": {
3830
+ "order": 0
3831
+ }
3832
+ },
3833
+ {
3834
+ "class": "nirs4all.operators.transforms.signal.Gaussian",
3835
+ "params": {
3836
+ "order": 0,
3837
+ "sigma": 2
3838
+ }
3839
+ }
3840
+ ]
3841
+ },
3842
+ {
3843
+ "_or_": [
3844
+ null,
3845
+ "nirs4all.operators.transforms.nirs.ASLSBaseline",
3846
+ "nirs4all.operators.transforms.signal.Detrend"
3847
+ ]
3848
+ },
3849
+ {
3850
+ "_or_": [
3851
+ null,
3852
+ "nirs4all.operators.transforms.orthogonalization.OSC",
3853
+ {
3854
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3855
+ "params": {
3856
+ "n_components": 2
3857
+ }
3858
+ },
3859
+ {
3860
+ "class": "nirs4all.operators.transforms.orthogonalization.OSC",
3861
+ "params": {
3862
+ "n_components": 3
3863
+ }
3864
+ }
3865
+ ]
3866
+ }
3867
+ ],
3868
+ "count": 90
3869
+ },
3870
+ {
3871
+ "class": "sklearn.model_selection._split.StratifiedKFold",
3872
+ "params": {
3873
+ "n_splits": 3,
3874
+ "shuffle": true,
3875
+ "random_state": 42
3876
+ }
3877
+ },
3878
+ {
3879
+ "class": "sklearn.preprocessing._data.StandardScaler",
3880
+ "params": {
3881
+ "with_std": false
3882
+ }
3883
+ },
3884
+ {
3885
+ "model": "nirs4all.operators.models.sklearn.plsda.PLSDA",
3886
+ "name": "PLSDA-Finetuned",
3887
+ "finetune_params": {
3888
+ "n_trials": 35,
3889
+ "sampler": "binary",
3890
+ "model_params": {
3891
+ "n_components": [
3892
+ "int",
3893
+ 1,
3894
+ 30
3895
+ ]
3896
+ }
3897
+ }
3898
+ },
3899
+ {
3900
+ "model": "sklearn.linear_model._ridge.RidgeClassifier",
3901
+ "name": "RidgeClassifier-Finetuned",
3902
+ "finetune_params": {
3903
+ "n_trials": 40,
3904
+ "sampler": "tpe",
3905
+ "model_params": {
3906
+ "alpha": [
3907
+ "log_float",
3908
+ 1e-05,
3909
+ 10000.0
3910
+ ],
3911
+ "fit_intercept": [
3912
+ "categorical",
3913
+ [
3914
+ true,
3915
+ false
3916
+ ]
3917
+ ]
3918
+ }
3919
+ }
3920
+ },
3921
+ {
3922
+ "model": {
3923
+ "class": "lightgbm.LGBMClassifier",
3924
+ "params": {
3925
+ "random_state": 42,
3926
+ "verbose": -1
3927
+ }
3928
+ },
3929
+ "name": "LightGBM-Finetuned",
3930
+ "finetune_params": {
3931
+ "n_trials": 30,
3932
+ "sampler": "tpe",
3933
+ "model_params": {
3934
+ "n_estimators": [
3935
+ "int",
3936
+ 100,
3937
+ 600
3938
+ ],
3939
+ "max_depth": [
3940
+ "int",
3941
+ 3,
3942
+ 12
3943
+ ],
3944
+ "learning_rate": [
3945
+ "log_float",
3946
+ 0.01,
3947
+ 0.2
3948
+ ],
3949
+ "num_leaves": [
3950
+ "int",
3951
+ 15,
3952
+ 127
3953
+ ],
3954
+ "subsample": [
3955
+ "float",
3956
+ 0.7,
3957
+ 1.0
3958
+ ]
3959
+ }
3960
+ }
3961
+ },
3962
+ {
3963
+ "model": "nirs4all.operators.models.sklearn.aom_pls_classifier.AOMPLSClassifier",
3964
+ "name": "AOM-PLS-Finetuned",
3965
+ "finetune_params": {
3966
+ "n_trials": 30,
3967
+ "sampler": "tpe",
3968
+ "model_params": {
3969
+ "n_components": [
3970
+ "int",
3971
+ 1,
3972
+ 25
3973
+ ]
3974
+ }
3975
+ }
3976
+ },
3977
+ {
3978
+ "model": {
3979
+ "class": "tabpfn.TabPFNClassifier",
3980
+ "params": {
3981
+ "ignore_pretraining_limits": true
3982
+ }
3983
+ },
3984
+ "name": "TabPFN"
3985
+ }
3986
+ ]
3987
+ }
3988
+ },
3989
+ "available_variants": [
3990
+ "regression",
3991
+ "classification"
3992
+ ],
3993
+ "steps_count": 8
3994
+ }
3995
+ ]
3996
+ }