Datasets:
Formats:
json
Languages:
English
Size:
< 1K
Tags:
time-series
time-series-decomposition
benchmark
component-recovery
symbolic-regression
icml-2026
License:
File size: 4,667 Bytes
17b7ba4 | 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 | """Benchmark configuration for TSComp v1.0.0 runbook."""
from __future__ import annotations
from typing import Dict, List, Sequence
BENCHMARK_VERSION = "1.0.0"
SUITES: Dict[str, List[str]] = {
"core": [
"trend_only_linear",
"trend_plus_single_sine",
"poly_trend_multi_harmonic",
"logistic_trend_multi_seasonal",
"rw_trend_freq_drifting_cycle",
"piecewise_trend_regime_cycle_with_events",
],
"full": [
"trend_only_linear",
"trend_plus_single_sine",
"poly_trend_multi_harmonic",
"logistic_trend_multi_seasonal",
"rw_trend_freq_drifting_cycle",
"piecewise_trend_regime_cycle_with_events",
],
}
SCENARIO_TIER: Dict[str, int] = {
"trend_only_linear": 1,
"trend_plus_single_sine": 1,
"poly_trend_multi_harmonic": 2,
"logistic_trend_multi_seasonal": 2,
"rw_trend_freq_drifting_cycle": 3,
"piecewise_trend_regime_cycle_with_events": 3,
}
# Protocol periods for injection (length=512, dt=1.0).
SCENARIO_PERIODS: Dict[str, List[int]] = {
"trend_only_linear": [50],
"trend_plus_single_sine": [50],
"poly_trend_multi_harmonic": [48],
"logistic_trend_multi_seasonal": [24, 168],
"rw_trend_freq_drifting_cycle": [50],
"piecewise_trend_regime_cycle_with_events": [40, 65],
}
CORE_METHODS: List[str] = [
"stl",
"mstl",
"ssa",
"emd",
"ceemdan",
"vmd",
"wavelet",
]
DEFAULT_METHOD_CONFIGS: Dict[str, Dict[str, object]] = {
"stl": {"period": None},
"mstl": {"periods": None},
"robuststl": {"period": None},
"ssa": {"window": None, "rank": 10, "primary_period": None},
"emd": {"primary_period": None},
"ceemdan": {"primary_period": None},
"vmd": {
"K": None, # v1.1.0: auto-calculated from periods (None triggers dynamic default)
"alpha": 300.0, # v1.1.0: reduced from 2000 for better mode separation
"tau": 0.0,
"DC": 0,
"init": 1,
"tol": 1e-7,
"seasonal_num_modes": 1,
"primary_period": None,
},
"wavelet": {"wavelet": "db4", "level": None},
"ma_baseline": {"trend_window": None, "season_period": None},
"dr_ts_reg": {
"lambda_T": 5.0, # v1.1.0: reduced from 100 to prevent over-smoothing
"lambda_S": 50.0,
"lambda_R": 0.1,
"period": None,
},
"dr_ts_ae": {
"model_path": None,
"latent_dim": 16,
"hidden_channels": [32, 64],
"kernel_size": 7,
"alpha_T": 10.0,
"alpha_S": 5.0,
"n_epochs": 50,
"device": "cpu",
"cache_model": True,
# WARNING: ORACLE method - trains on test data
},
"sl_lib": {
"library_size": 500,
"n_trend_bases": 200,
"n_seasonal_bases": 300,
"n_candidates": 100, # v1.1.0: increased from 50
"sparsity_lambda": 0.001, # v1.1.0: reduced from 0.01
"max_poly_degree": 5,
"min_period": 4,
"max_period": 128,
},
}
def list_suites() -> List[str]:
return sorted(SUITES.keys())
def get_suite(suite: str) -> List[str]:
key = suite.strip().lower()
if key not in SUITES:
raise ValueError(f"Unknown suite '{suite}'. Available: {list_suites()}")
return list(SUITES[key])
def normalize_methods(methods: Sequence[str]) -> List[str]:
return [m.strip().lower() for m in methods if m and m.strip()]
def resolve_methods(methods: str | Sequence[str]) -> List[str]:
if isinstance(methods, str):
raw = methods.strip().lower()
if raw in {"core", "official"}:
return list(CORE_METHODS)
if raw in {"all", "full"}:
return sorted(DEFAULT_METHOD_CONFIGS.keys())
return normalize_methods(raw.split(","))
return normalize_methods(methods)
def normalize_periods(periods: Sequence[int], length: int) -> List[int]:
cleaned: List[int] = []
max_period = max(2, length // 2)
for val in periods:
try:
p = int(round(float(val)))
except (TypeError, ValueError):
continue
p = max(2, min(p, max_period))
if p not in cleaned:
cleaned.append(p)
return cleaned
def select_primary_period(periods: Sequence[int]) -> int | None:
for val in periods:
if val is None:
continue
try:
p = int(round(float(val)))
except (TypeError, ValueError):
continue
if p >= 2:
return p
return None
|