Datasets:
Formats:
json
Languages:
English
Size:
< 1K
Tags:
time-series
time-series-decomposition
benchmark
component-recovery
symbolic-regression
icml-2026
License:
File size: 12,540 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 | """Plotting helpers for comparing decomposition methods."""
from __future__ import annotations
import math
from pathlib import Path
from typing import Any, Dict, Iterable, List, Optional, Sequence
import matplotlib.pyplot as plt
import numpy as np
from .decomp_methods import (
DecompConfig,
DecompMethodName,
DecompResult,
decompose_series,
)
METHOD_COLORS: Dict[str, str] = {
"stl": "tab:blue",
"mstl": "tab:orange",
"robuststl": "tab:red",
"ssa": "tab:green",
"std": "tab:olive",
"emd": "tab:purple",
"ceemdan": "tab:brown",
"vmd": "tab:pink",
"wavelet": "tab:cyan",
"ma_baseline": "tab:gray",
}
def _method_color(method: str) -> str:
return METHOD_COLORS.get(method.lower(), "tab:gray")
def _chunk_list(values: Sequence[str], size: int) -> Iterable[List[str]]:
size = max(1, size)
for i in range(0, len(values), size):
yield list(values[i : i + size])
def _component_ylim(arrays: List[np.ndarray]) -> Optional[tuple[float, float]]:
filtered = [np.asarray(arr, dtype=float).ravel() for arr in arrays if arr is not None and np.asarray(arr).size]
if not filtered:
return None
data = np.concatenate(filtered)
lo, hi = float(np.min(data)), float(np.max(data))
pad = 0.05 * (hi - lo) if hi > lo else 1.0
return lo - pad, hi + pad
def compare_decompositions_on_series(
y: np.ndarray,
methods: List[Dict[str, Any]],
title: str = "",
fs: float = 1.0,
meta: Optional[Dict[str, Any]] = None,
) -> plt.Figure:
"""
Plot a side-by-side comparison of multiple decompositions on the same series.
"""
series = np.asarray(y, dtype=float).reshape(-1)
if not methods:
raise ValueError("Provide at least one method to compare.")
results: List[Dict[str, Any]] = []
for entry in methods:
name = entry.get("name")
if name is None:
raise ValueError("Each method entry must include a 'name' key.")
config = entry.get("config")
label = entry.get("label", name)
res = decompose_series(series, method=name, config=config, fs=fs, meta=meta)
results.append({"label": label, "result": res})
fig, axes = plt.subplots(4, 1, sharex=True, figsize=(10, 8))
axes[0].plot(series, color="black", label="observed")
axes[0].set_title("Observed series")
axes[0].legend(loc="upper right")
axes[1].set_title("Trend components")
axes[2].set_title("Seasonal components")
axes[3].set_title("Residual components")
for item in results:
label = item["label"]
res: DecompResult = item["result"]
axes[1].plot(res.trend, label=label)
axes[2].plot(res.season, label=label)
axes[3].plot(res.residual, label=label)
for ax in axes[1:]:
ax.legend(loc="upper right")
if title:
fig.suptitle(title)
axes[-1].set_xlabel("Time index")
plt.tight_layout()
return fig
def plot_decomposition_overlays_paginated(
scenario_name: str,
observed: np.ndarray,
true_components: Dict[str, np.ndarray],
components_by_method: Dict[str, Dict[str, np.ndarray]],
output_dir: str | Path,
methods: Optional[List[str]] = None,
max_methods_per_figure: int = 4,
) -> List[Path]:
"""
Create paginated overlay plots (observed + T/S/R) for a subset of methods.
"""
output_dir = Path(output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
methods = methods or sorted(components_by_method.keys())
saved_paths: List[Path] = []
slug = scenario_name.replace(" ", "_")
for page_idx, chunk in enumerate(_chunk_list(methods, max_methods_per_figure), start=1):
fig, axes = plt.subplots(4, 1, sharex=True, figsize=(11, 8))
axes[0].plot(observed, color="black", label="observed")
axes[0].set_title("Observed series")
axes[0].legend(loc="upper right")
components = ["trend", "seasonal", "residual"]
titles = ["Trend components", "Seasonal components", "Residual components"]
for ax, comp, title in zip(axes[1:], components, titles):
ax.set_title(title)
true_comp = true_components.get(comp)
if true_comp is not None:
ax.plot(true_comp, color="black", linewidth=1.6, label="true")
for method in chunk:
result = components_by_method.get(method)
if not result:
continue
color = _method_color(method)
for ax, comp in zip(axes[1:], components):
comp_data = result.get(comp)
if comp_data is not None:
ax.plot(comp_data, color=color, linewidth=1.0, alpha=0.9, label=method)
axes[1].legend(loc="upper right", ncol=1, fontsize="small")
axes[-1].set_xlabel("Time index")
fig.suptitle(f"Scenario: {scenario_name} – Methods {', '.join(chunk)}", fontsize=14)
fig.tight_layout(rect=[0, 0, 1, 0.97])
out_path = output_dir / f"decomp_overlay_{slug}_page{page_idx}.png"
fig.savefig(out_path, dpi=150)
plt.close(fig)
saved_paths.append(out_path)
return saved_paths
def plot_decomposition_facets(
scenario_name: str,
observed: np.ndarray,
true_components: Dict[str, np.ndarray],
components_by_method: Dict[str, Dict[str, np.ndarray]],
output_dir: str | Path,
methods: Optional[List[str]] = None,
max_methods_per_page: int = 4,
) -> List[Path]:
"""
Create per-method facet plots (observed row + method-specific T/S/R rows).
"""
output_dir = Path(output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
methods = methods or sorted(components_by_method.keys())
saved_paths: List[Path] = []
slug = scenario_name.replace(" ", "_")
for page_idx, chunk in enumerate(_chunk_list(methods, max_methods_per_page), start=1):
n_methods = len(chunk)
n_rows = 1 + 3 * n_methods
fig, axes = plt.subplots(n_rows, 1, sharex=True, figsize=(12, 3 + 2.4 * n_methods))
axes[0].plot(observed, color="black")
axes[0].set_title("Observed series")
trend_arrays = [components_by_method[m].get("trend") for m in chunk if m in components_by_method]
season_arrays = [components_by_method[m].get("seasonal") for m in chunk if m in components_by_method]
resid_arrays = [components_by_method[m].get("residual") for m in chunk if m in components_by_method]
trend_ylim = _component_ylim([true_components.get("trend")] + trend_arrays)
season_ylim = _component_ylim([true_components.get("seasonal")] + season_arrays)
resid_ylim = _component_ylim([true_components.get("residual")] + resid_arrays)
for idx, method in enumerate(chunk):
base_idx = 1 + idx * 3
result = components_by_method.get(method)
if not result:
continue
color = _method_color(method)
# Trend
ax_trend = axes[base_idx]
if true_components.get("trend") is not None:
ax_trend.plot(true_components["trend"], color="black", linewidth=1.5, label="true")
ax_trend.plot(result.get("trend"), color=color, label=method)
ax_trend.set_ylabel(method)
ax_trend.set_title(f"Trend – {method}")
if trend_ylim:
ax_trend.set_ylim(trend_ylim)
# Seasonal
ax_season = axes[base_idx + 1]
if true_components.get("seasonal") is not None:
ax_season.plot(true_components["seasonal"], color="black", linewidth=1.5)
ax_season.plot(result.get("seasonal"), color=color)
ax_season.set_title(f"Seasonal – {method}")
if season_ylim:
ax_season.set_ylim(season_ylim)
# Residual
ax_resid = axes[base_idx + 2]
if true_components.get("residual") is not None:
ax_resid.plot(true_components["residual"], color="black", linewidth=1.2)
ax_resid.plot(result.get("residual"), color=color)
ax_resid.set_title(f"Residual – {method}")
if resid_ylim:
ax_resid.set_ylim(resid_ylim)
axes[-1].set_xlabel("Time index")
fig.suptitle(f"Scenario: {scenario_name} – methods {', '.join(chunk)}", fontsize=14)
fig.tight_layout(rect=[0, 0, 1, 0.97])
out_path = output_dir / f"decomp_facets_{slug}_page{page_idx}.png"
fig.savefig(out_path, dpi=150)
plt.close(fig)
saved_paths.append(out_path)
return saved_paths
def plot_component_error_timeseries(
scenario_name: str,
true_components: Dict[str, np.ndarray],
components_by_method: Dict[str, Dict[str, np.ndarray]],
output_dir: str | Path,
component: str = "trend",
methods_to_show: Optional[List[str]] = None,
) -> Path:
"""
Plot absolute error over time for a given component across methods.
"""
output_dir = Path(output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
methods = methods_to_show or sorted(components_by_method.keys())
true_array = true_components.get(component)
if true_array is None:
raise ValueError(f"No ground-truth component '{component}' available.")
true_series = np.asarray(true_array, dtype=float).reshape(-1)
fig, ax = plt.subplots(figsize=(12, 4))
for method in methods:
result = components_by_method.get(method)
if not result or component not in result:
continue
est = np.asarray(result[component], dtype=float).reshape(-1)
length = min(len(est), len(true_series))
if length == 0:
continue
abs_err = np.abs(est[:length] - true_series[:length])
ax.plot(abs_err, label=method, color=_method_color(method))
ax.set_title(f"{scenario_name} – {component} absolute error over time")
ax.set_xlabel("Time index")
ax.set_ylabel("|estimate - true|")
ax.legend(loc="upper right", ncol=2, fontsize="small")
fig.tight_layout()
out_path = output_dir / f"error_timeseries_{component}_{scenario_name.replace(' ', '_')}.png"
fig.savefig(out_path, dpi=150)
plt.close(fig)
return out_path
def plot_component_error_heatmap(
scenario_name: str,
true_components: Dict[str, np.ndarray],
components_by_method: Dict[str, Dict[str, np.ndarray]],
output_dir: str | Path,
component: str = "trend",
methods_to_show: Optional[List[str]] = None,
) -> Path:
"""
Plot heatmap of absolute errors over time for multiple methods.
"""
output_dir = Path(output_dir)
output_dir.mkdir(parents=True, exist_ok=True)
methods = methods_to_show or sorted(components_by_method.keys())
true_array = true_components.get(component)
if true_array is None:
raise ValueError(f"No ground-truth component '{component}' available.")
true_series = np.asarray(true_array, dtype=float).reshape(-1)
error_matrix = []
valid_methods = []
for method in methods:
result = components_by_method.get(method)
if not result or component not in result:
continue
est = np.asarray(result[component], dtype=float).reshape(-1)
length = min(len(est), len(true_series))
if length == 0:
continue
abs_err = np.abs(est[:length] - true_series[:length])
error_matrix.append(abs_err)
valid_methods.append(method)
if not error_matrix:
raise ValueError("No valid error series to plot.")
data = np.vstack(error_matrix)
fig, ax = plt.subplots(figsize=(12, 0.5 * len(valid_methods) + 2))
im = ax.imshow(data, aspect="auto", interpolation="nearest", cmap="viridis")
ax.set_title(f"{scenario_name} – {component} absolute error heatmap")
ax.set_xlabel("Time index")
ax.set_yticks(range(len(valid_methods)))
ax.set_yticklabels(valid_methods)
fig.colorbar(im, ax=ax, shrink=0.75, label="abs error")
fig.tight_layout()
out_path = output_dir / f"error_heatmap_{component}_{scenario_name.replace(' ', '_')}.png"
fig.savefig(out_path, dpi=150)
plt.close(fig)
return out_path
|