File size: 11,560 Bytes
ae97a0a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
{
  "recommendations": [
    {
      "metadata": {
        "assumptions": [
          "expects a 2D array with at least two aligned channels",
          "works best when window and rank reflect the dominant temporal structure",
          "MSSA should be evaluated against residual diagnostics rather than used as a black box"
        ],
        "dependency_tier": "core",
        "example_config": {
          "backend": "python",
          "channel_names": [
            "channel_a",
            "channel_b",
            "channel_c"
          ],
          "method": "MSSA",
          "params": {
            "primary_period": 12,
            "rank": 6,
            "window": 24
          },
          "speed_mode": "exact"
        },
        "family": "SSA",
        "implementation": "python",
        "input_mode": "multivariate",
        "maturity": "flagship",
        "min_length": 24,
        "multivariate_support": "shared-model",
        "name": "MSSA",
        "native_backed": false,
        "not_recommended_for": [
          "single-series workflows where a univariate flagship method is sufficient",
          "very short series that cannot support a sensible window length"
        ],
        "optional_dependencies": [],
        "output_components": [
          "trend",
          "season",
          "residual",
          "components.elementary"
        ],
        "package_links": [
          {
            "note": "SSA-focused package; useful comparison point for SSA-family workflows.",
            "title": "SSALib",
            "url": "https://github.com/ADSCIAN/ssalib"
          }
        ],
        "parameter_docs": [
          {
            "common": true,
            "default": null,
            "description": "Shared embedding window length for aligned channels.",
            "name": "window",
            "required": true,
            "type": "int"
          },
          {
            "common": true,
            "default": null,
            "description": "Number of shared elementary components to retain.",
            "name": "rank",
            "required": false,
            "type": "int | None"
          },
          {
            "common": true,
            "default": null,
            "description": "Dominant shared period used by automatic grouping.",
            "name": "primary_period",
            "required": false,
            "type": "int | None"
          },
          {
            "common": false,
            "default": 1.0,
            "description": "Sampling frequency used by frequency-based grouping.",
            "name": "fs",
            "required": false,
            "type": "float"
          },
          {
            "common": false,
            "default": null,
            "description": "Explicit component indexes assigned to trend.",
            "name": "trend_components",
            "required": false,
            "type": "list[int] | None"
          },
          {
            "common": false,
            "default": null,
            "description": "Explicit component indexes assigned to season.",
            "name": "season_components",
            "required": false,
            "type": "list[int] | None"
          }
        ],
        "recommended_for": [
          "multivariate component recovery",
          "shared seasonal structure across channels",
          "accuracy-first multivariate workflows"
        ],
        "references": [
          {
            "note": "Primary SSA/MSSA reference used for the multivariate extension.",
            "title": "Golyandina and Zhigljavsky (2020), Singular Spectrum Analysis for Time Series",
            "url": "https://link.springer.com/book/10.1007/978-3-662-62436-4"
          }
        ],
        "summary": "Multivariate SSA for shared-structure decomposition across channels.",
        "typical_failure_modes": [
          "too few channels for MSSA",
          "window or rank too small for the shared structure"
        ]
      },
      "method": "MSSA",
      "rank": 1,
      "reason_codes": [
        "maturity:flagship",
        "shared_multivariate",
        "accuracy_shortlist",
        "multivariate_accuracy_bonus"
      ],
      "score": 13.5,
      "summary": "Multivariate SSA for shared-structure decomposition across channels."
    },
    {
      "metadata": {
        "assumptions": [
          "treats each channel independently under one shared method surface",
          "works best when one seasonal period or block structure is reasonably stable",
          "STDR should be evaluated against residual diagnostics rather than used as a black box"
        ],
        "dependency_tier": "core",
        "example_config": {
          "backend": "auto",
          "method": "STDR",
          "params": {
            "period": 12
          },
          "speed_mode": "exact"
        },
        "family": "SeasonalTrend",
        "implementation": "native-backed",
        "input_mode": "channelwise",
        "maturity": "flagship",
        "min_length": 8,
        "multivariate_support": "channelwise",
        "name": "STDR",
        "native_backed": true,
        "not_recommended_for": [
          "problems that require one shared latent model across channels",
          "series where the dominant period is unknown and cannot be inferred reliably"
        ],
        "optional_dependencies": [],
        "output_components": [
          "trend",
          "season",
          "residual",
          "components.dispersion",
          "components.seasonal_shape"
        ],
        "package_links": [],
        "parameter_docs": [
          {
            "common": true,
            "default": null,
            "description": "Seasonal period in samples.",
            "name": "period",
            "required": true,
            "type": "int"
          },
          {
            "common": false,
            "default": null,
            "description": "Optional search horizon when period inference is enabled.",
            "name": "max_period_search",
            "required": false,
            "type": "int | None"
          },
          {
            "common": false,
            "default": 1e-08,
            "description": "Small numerical guard for robust dispersion calculations.",
            "name": "eps",
            "required": false,
            "type": "float"
          }
        ],
        "recommended_for": [
          "robust seasonal-trend decomposition",
          "channelwise multivariate workflows",
          "native-backed seasonal structure recovery"
        ],
        "references": [
          {
            "note": "Primary reference for STD and the robust seasonal-trend-dispersion family.",
            "title": "Dudek (2022), STD: A Seasonal-Trend-Dispersion Decomposition of Time Series",
            "url": "https://doi.org/10.48550/arXiv.2204.10398"
          }
        ],
        "summary": "Robust seasonal-trend decomposition for noisier periodic signals.",
        "typical_failure_modes": [
          "period omitted or mis-specified",
          "heavy structural breaks that violate shared seasonal assumptions"
        ]
      },
      "method": "STDR",
      "rank": 2,
      "reason_codes": [
        "maturity:flagship",
        "native_backed",
        "channelwise_multivariate",
        "accuracy_shortlist",
        "long_series_native"
      ],
      "score": 11.75,
      "summary": "Robust seasonal-trend decomposition for noisier periodic signals."
    },
    {
      "metadata": {
        "assumptions": [
          "treats each channel independently under one shared method surface",
          "works best when one seasonal period or block structure is reasonably stable",
          "STD should be evaluated against residual diagnostics rather than used as a black box"
        ],
        "dependency_tier": "core",
        "example_config": {
          "backend": "auto",
          "method": "STD",
          "params": {
            "period": 12
          },
          "speed_mode": "exact"
        },
        "family": "SeasonalTrend",
        "implementation": "native-backed",
        "input_mode": "channelwise",
        "maturity": "flagship",
        "min_length": 8,
        "multivariate_support": "channelwise",
        "name": "STD",
        "native_backed": true,
        "not_recommended_for": [
          "problems that require one shared latent model across channels",
          "series where the dominant period is unknown and cannot be inferred reliably"
        ],
        "optional_dependencies": [],
        "output_components": [
          "trend",
          "season",
          "residual",
          "components.dispersion",
          "components.seasonal_shape"
        ],
        "package_links": [],
        "parameter_docs": [
          {
            "common": true,
            "default": null,
            "description": "Seasonal period in samples.",
            "name": "period",
            "required": true,
            "type": "int"
          },
          {
            "common": false,
            "default": null,
            "description": "Optional search horizon when period inference is enabled.",
            "name": "max_period_search",
            "required": false,
            "type": "int | None"
          },
          {
            "common": false,
            "default": 1e-08,
            "description": "Small numerical guard for dispersion calculations.",
            "name": "eps",
            "required": false,
            "type": "float"
          }
        ],
        "recommended_for": [
          "fast seasonal-trend baselines",
          "channelwise multivariate workflows",
          "native-backed production paths"
        ],
        "references": [
          {
            "note": "Primary reference for STD and the robust seasonal-trend-dispersion family.",
            "title": "Dudek (2022), STD: A Seasonal-Trend-Dispersion Decomposition of Time Series",
            "url": "https://doi.org/10.48550/arXiv.2204.10398"
          }
        ],
        "summary": "Fast seasonal-trend decomposition with dispersion-aware diagnostics.",
        "typical_failure_modes": [
          "period omitted or mis-specified",
          "shared seasonal structure changing too quickly across cycles"
        ]
      },
      "method": "STD",
      "rank": 3,
      "reason_codes": [
        "maturity:flagship",
        "native_backed",
        "channelwise_multivariate",
        "long_series_native"
      ],
      "score": 9.25,
      "summary": "Fast seasonal-trend decomposition with dispersion-aware diagnostics."
    }
  ],
  "rejected_methods": {
    "CEEMDAN": "univariate_only",
    "EMD": "univariate_only",
    "GABOR_CLUSTER": "univariate_only",
    "MA_BASELINE": "univariate_only",
    "MEMD": "optional_backend_disabled",
    "MSTL": "univariate_only",
    "MVMD": "optional_backend_disabled",
    "ROBUST_STL": "univariate_only",
    "SSA": "univariate_only",
    "STL": "univariate_only",
    "VMD": "univariate_only",
    "WAVELET": "univariate_only"
  },
  "request": {
    "allow_optional_backends": false,
    "channels": 3,
    "length": 192,
    "prefer": "accuracy",
    "require_native": false,
    "top_k": 5
  }
}