Graphiques de corrélations binaires simples

docs/03 - Premiers graphiques/index.md

@@ -4,27 +4,29 @@ On peut désormais tracer nos premiers graphiques simples et bruts.
 S'ils ne sont pas très instructifs par rapport à ce que nous fournissent Home Assistant et InfluxDB, ils nous permettent au moins de nous assurer que tout fonctionne, et que les données semblent cohérentes.
 Les fichiers CSV correspondant à chaque figure sont conservés dans `data/` dans ce dossier.
 
-On se limite dans un premier temps aux 7 derniers jours.
+Les graphiques couvrent maintenant toute la période disponible dans `data/weather_minutely.csv`.
+Une agrégation automatique réduit le nombre de points pour rester lisible (plus de courbes "peignes"), et l'axe des dates utilise un format compact qui évite tout chevauchement de labels.
+On peut au besoin restreindre la période avec `--days` ou imposer une fréquence d'agrégation avec `--resample`.
 
 ```shell
 python "docs/03 - Premiers graphiques/scripts/plot_basic_variables.py"
 ```
 
-![](figures/temperature_last_7_days.png)
+![](figures/temperature_overview.png)
 
-![](figures/pressure_last_7_days.png)
+![](figures/pressure_overview.png)
 
-![](figures/humidity_last_7_days.png)
+![](figures/humidity_overview.png)
 
-![](figures/rain_rate_last_7_days.png)
+![](figures/rain_rate_overview.png)
 
-![](figures/illuminance_last_7_days.png)
+![](figures/wind_speed_overview.png)
 
-![](figures/wind_speed_last_7_days.png)
+![](figures/wind_direction_overview.png)
 
-![](figures/wind_direction_last_7_days.png)
+![](figures/illuminance_overview.png)
 
-![](figures/sun_elevation_last_7_days.png)
+![](figures/sun_elevation_overview.png)
 
 ## Vues calendrier
 

docs/03 - Premiers graphiques/scripts/plot_basic_variables.py

@@ -1,5 +1,5 @@
 # scripts/plot_basic_variables.py
-"""Génère des séries temporelles simples (7 jours) pour chaque variable météo."""
+"""Génère des séries temporelles simples pour chaque variable météo."""
 
 from __future__ import annotations
 
@@ -7,7 +7,6 @@ import argparse
 from pathlib import Path
 import sys
 
-import matplotlib.pyplot as plt
 import pandas as pd
 
 
@@ -16,7 +15,7 @@ if str(PROJECT_ROOT) not in sys.path:
     sys.path.insert(0, str(PROJECT_ROOT))
 
 from meteo.dataset import load_raw_csv
-from meteo.plots import export_plot_dataset
+from meteo.plots import PlotChoice, PlotStyle, plot_basic_series, recommended_style, resample_series_for_plot
 from meteo.variables import Variable, VARIABLES
 
 
@@ -25,47 +24,32 @@ DOC_DIR = Path(__file__).resolve().parent.parent
 DEFAULT_OUTPUT_DIR = DOC_DIR / "figures"
 
 
-def _prepare_slice(df: pd.DataFrame, *, last_days: int) -> pd.DataFrame:
+def _select_window(df: pd.DataFrame, *, last_days: int | None) -> pd.DataFrame:
-    """Extrait la fenêtre temporelle souhaitée et applique une moyenne horaire pour lisser la courbe."""
+    """Extrait la fenêtre temporelle souhaitée (ou la totalité si None)."""
 
+    if last_days is None:
+        return df
     end = df.index.max()
     start = end - pd.Timedelta(days=last_days)
-    df_slice = df.loc[start:end]
+    return df.loc[start:end]
-    numeric_slice = df_slice.select_dtypes(include="number")
-    if numeric_slice.empty:
-        raise RuntimeError("Aucune colonne numérique disponible pour les moyennes horaires.")
-    return numeric_slice.resample("1h").mean()
 
 
-def _plot_variable(df_hourly: pd.DataFrame, var: Variable, output_dir: Path) -> Path | None:
+def _format_ylabel(var: Variable) -> str:
-    """Trace la série pour une variable et retourne le chemin de l'image générée."""
-
-    if var.column not in df_hourly.columns:
-        print(f"⚠ Colonne absente pour {var.key} ({var.column}).")
-        return None
-
-    series = df_hourly[var.column].dropna()
-    if series.empty:
-        print(f"⚠ Aucun point valide pour {var.key} dans l'intervalle choisi.")
-        return None
-
-    output_dir.mkdir(parents=True, exist_ok=True)
-    output_path = output_dir / f"{var.key}_last_7_days.png"
-
-    export_plot_dataset(series.to_frame(name=var.column), output_path)
-
-    plt.figure()
-    plt.plot(series.index, series)
-    plt.xlabel("Temps (UTC)")
     unit_text = f" ({var.unit})" if var.unit else ""
-    plt.ylabel(f"{var.label}{unit_text}")
+    return f"{var.label}{unit_text}"
-    plt.title(f"{var.label} - Moyenne horaire sur les 7 derniers jours")
+
-    plt.grid(True)
+
-    plt.tight_layout()
+def _aggregation_label(choice: PlotChoice, freq: str) -> str:
-    plt.savefig(output_path, dpi=150)
+    """Texte court pour indiquer l'agrégation appliquée."""
-    plt.close()
+
-    print(f"✔ Graphique généré : {output_path}")
+    base = "moyenne"
-    return output_path
+    if callable(choice.agg) and getattr(choice.agg, "__name__", "") == "_circular_mean_deg":
+        base = "moyenne circulaire"
+    elif choice.agg == "sum":
+        base = "somme"
+    elif choice.agg == "median":
+        base = "médiane"
+    return f"{base} {freq}"
 
 
 def main(argv: list[str] | None = None) -> None:
@@ -78,8 +62,23 @@ def main(argv: list[str] | None = None) -> None:
     parser.add_argument(
         "--days",
         type=int,
-        default=7,
+        default=None,
-        help="Nombre de jours à afficher (par défaut : 7).",
+        help="Nombre de jours à afficher (par défaut : toute la période disponible).",
+    )
+    parser.add_argument(
+        "--style",
+        choices=[style.value for style in PlotStyle],
+        help="Style de représentation à utiliser pour toutes les variables (par défaut : recommandations par variable).",
+    )
+    parser.add_argument(
+        "--resample",
+        help="Fréquence pandas à utiliser pour l'agrégation temporelle (par défaut : calcul automatique).",
+    )
+    parser.add_argument(
+        "--max-points",
+        type=int,
+        default=420,
+        help="Nombre de points cible après agrégation automatique (par défaut : 420).",
     )
     parser.add_argument(
         "--output-dir",
@@ -93,7 +92,7 @@ def main(argv: list[str] | None = None) -> None:
         raise FileNotFoundError(f"Dataset introuvable : {CSV_PATH}")
 
     df = load_raw_csv(CSV_PATH)
-    df_hourly = _prepare_slice(df, last_days=args.days)
+    df_window = _select_window(df, last_days=args.days)
 
     selected: list[Variable]
     if args.only:
@@ -105,8 +104,44 @@ def main(argv: list[str] | None = None) -> None:
     else:
         selected = list(VARIABLES)
 
+    output_dir: Path = args.output_dir
+    output_dir.mkdir(parents=True, exist_ok=True)
+
     for variable in selected:
-        _plot_variable(df_hourly, variable, args.output_dir)
+        if variable.column not in df_window.columns:
+            print(f"⚠ Colonne absente pour {variable.key} ({variable.column}).")
+            continue
+
+        series = df_window[variable.column].dropna()
+        if series.empty:
+            print(f"⚠ Aucun point valide pour {variable.key} sur la période choisie.")
+            continue
+
+        style_choice = recommended_style(variable, args.style)
+
+        aggregated, freq_used = resample_series_for_plot(
+            series,
+            variable=variable,
+            freq=args.resample,
+            target_points=args.max_points,
+        )
+        if aggregated.empty:
+            print(f"⚠ Pas de points après agrégation pour {variable.key}.")
+            continue
+
+        output_path = output_dir / f"{variable.key}_overview.png"
+        annotate_freq = _aggregation_label(style_choice, freq_used)
+
+        plot_basic_series(
+            aggregated,
+            variable=variable,
+            output_path=output_path,
+            style=style_choice.style,
+            title=f"{variable.label} — évolution temporelle",
+            ylabel=_format_ylabel(variable),
+            annotate_freq=annotate_freq,
+        )
+        print(f"✔ Graphique généré : {output_path}")
 
 
 if __name__ == "__main__":

docs/04 - Corrélations binaires/index.md

@@ -1,7 +1,66 @@
 # Corrélations binaires
 
-Cette étape regroupe l'ensemble des scripts dédiés aux corrélations et comparaisons directes entre variables.
+## Superpositions simples
-Chaque figure déposée dans `figures/` possède son CSV compagnon exporté dans le dossier `data/` au même emplacement.
+
+```shell
+python "docs/04 - Corrélations binaires/scripts/plot_pairwise_time_series.py"
+```
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_humidity.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_pressure.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_rain_rate.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_illuminance.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_wind_speed.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_temperature_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_pressure.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_rain_rate.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_illuminance.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_wind_speed.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_humidity_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_pressure_vs_rain_rate.png)
+
+![](figures/pairwise_timeseries/timeseries_pressure_vs_illuminance.png)
+
+![](figures/pairwise_timeseries/timeseries_pressure_vs_wind_speed.png)
+
+![](figures/pairwise_timeseries/timeseries_pressure_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_pressure_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_rain_rate_vs_wind_speed.png)
+
+![](figures/pairwise_timeseries/timeseries_rain_rate_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_rain_rate_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_illuminance_vs_wind_speed.png)
+
+![](figures/pairwise_timeseries/timeseries_illuminance_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_illuminance_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_wind_speed_vs_wind_direction.png)
+
+![](figures/pairwise_timeseries/timeseries_wind_speed_vs_sun_elevation.png)
+
+![](figures/pairwise_timeseries/timeseries_wind_direction_vs_sun_elevation.png)
+
+## Nuages de points
 
 ```shell
 python "docs/04 - Corrélations binaires/scripts/plot_all_pairwise_scatter.py"

docs/04 - Corrélations binaires/scripts/plot_pairwise_time_series.py

@@ -0,0 +1,169 @@
+# scripts/plot_pairwise_time_series.py
+from __future__ import annotations
+
+from pathlib import Path
+import sys
+
+import argparse
+import pandas as pd
+
+PROJECT_ROOT = Path(__file__).resolve().parents[3]
+if str(PROJECT_ROOT) not in sys.path:
+    sys.path.insert(0, str(PROJECT_ROOT))
+
+from meteo.dataset import load_raw_csv
+from meteo.plots import (
+    PlotChoice,
+    PlotStyle,
+    plot_dual_time_series,
+    recommended_style,
+    resample_series_for_plot,
+)
+from meteo.variables import Variable, VARIABLES, VARIABLES_BY_KEY, iter_variable_pairs
+
+
+CSV_PATH = Path("data/weather_minutely.csv")
+DOC_DIR = Path(__file__).resolve().parent.parent
+OUTPUT_DIR = DOC_DIR / "figures" / "pairwise_timeseries"
+
+
+def _select_variables(keys: list[str] | None) -> list[Variable]:
+    if not keys:
+        return list(VARIABLES)
+    missing = [key for key in keys if key not in VARIABLES_BY_KEY]
+    if missing:
+        raise KeyError(f"Variables inconnues : {', '.join(missing)}")
+    return [VARIABLES_BY_KEY[key] for key in keys]
+
+
+def _aggregation_label(choice_a: PlotChoice, choice_b: PlotChoice, freq: str) -> str:
+    agg_labels = set()
+    for choice in (choice_a, choice_b):
+        base = "moyenne"
+        if isinstance(choice.agg, str):
+            if choice.agg == "sum":
+                base = "somme"
+            elif choice.agg == "median":
+                base = "médiane"
+        elif getattr(choice.agg, "__name__", "") == "_circular_mean_deg":
+            base = "moyenne circulaire"
+        agg_labels.add(base)
+    if len(agg_labels) == 1:
+        label = agg_labels.pop()
+    else:
+        label = "agrégations mixtes"
+    return f"{label} {freq}"
+
+
+def main(argv: list[str] | None = None) -> None:
+    parser = argparse.ArgumentParser(description="Superpose les séries temporelles de toutes les paires de variables.")
+    parser.add_argument(
+        "--only",
+        nargs="*",
+        help="Clés de variables à inclure (par défaut : toutes).",
+    )
+    parser.add_argument(
+        "--days",
+        type=int,
+        default=None,
+        help="Limiter aux N derniers jours (par défaut : période complète).",
+    )
+    parser.add_argument(
+        "--style",
+        choices=[style.value for style in PlotStyle],
+        help="Style à imposer à toutes les variables (par défaut : style recommandé par variable).",
+    )
+    parser.add_argument(
+        "--resample",
+        help="Fréquence pandas pour l'agrégation temporelle (par défaut : calcul automatique).",
+    )
+    parser.add_argument(
+        "--max-points",
+        type=int,
+        default=420,
+        help="Nombre de points cible après agrégation automatique (par défaut : 420).",
+    )
+    parser.add_argument(
+        "--output-dir",
+        type=Path,
+        default=OUTPUT_DIR,
+        help="Dossier où stocker les figures.",
+    )
+    args = parser.parse_args(argv)
+
+    if not CSV_PATH.exists():
+        print(f"⚠ Fichier introuvable : {CSV_PATH}")
+        return
+
+    df = load_raw_csv(CSV_PATH)
+    if args.days is not None:
+        end = df.index.max()
+        start = end - pd.Timedelta(days=args.days)
+        df = df.loc[start:end]
+
+    variables = _select_variables(args.only)
+    pairs = [(vx, vy) for (vx, vy) in iter_variable_pairs() if vx in variables and vy in variables]
+    if not pairs:
+        print("⚠ Aucune paire à tracer.")
+        return
+
+    output_dir: Path = args.output_dir
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    print(f"Dataset chargé ({len(df)} lignes) → génération de {len(pairs)} paires.")
+
+    for var_a, var_b in pairs:
+        missing: list[str] = []
+        for col in (var_a.column, var_b.column):
+            if col not in df.columns:
+                missing.append(col)
+        if missing:
+            print(f"⚠ Colonnes absentes, on passe : {', '.join(missing)}")
+            continue
+
+        series_a = df[var_a.column].dropna()
+        series_b = df[var_b.column].dropna()
+        if series_a.empty or series_b.empty:
+            print(f"⚠ Données insuffisantes pour {var_a.key} / {var_b.key}, on passe.")
+            continue
+
+        choice_a = recommended_style(var_a, args.style)
+        choice_b = recommended_style(var_b, args.style)
+
+        aggregated_a, freq_used = resample_series_for_plot(
+            series_a,
+            variable=var_a,
+            freq=args.resample,
+            target_points=args.max_points,
+        )
+        aggregated_b, _ = resample_series_for_plot(
+            series_b,
+            variable=var_b,
+            freq=freq_used,
+            target_points=args.max_points,
+        )
+        if aggregated_a.empty or aggregated_b.empty:
+            print(f"⚠ Pas de points après agrégation pour {var_a.key} / {var_b.key}.")
+            continue
+
+        output_path = output_dir / f"timeseries_{var_a.key}_vs_{var_b.key}.png"
+        label_freq = _aggregation_label(choice_a, choice_b, freq_used)
+
+        print(f"→ {var_a.key} vs {var_b.key} ({freq_used}) → {output_path}")
+        plot_dual_time_series(
+            aggregated_a,
+            var_a,
+            choice_a,
+            aggregated_b,
+            var_b,
+            choice_b,
+            output_path=output_path,
+            title=f"{var_a.label} et {var_b.label} — évolution temporelle",
+            annotate_freq=label_freq,
+        )
+
+    print("✔ Superpositions temporelles générées.")
+
+
+if __name__ == "__main__":
+    main()

meteo/plots/__init__.py

@@ -22,6 +22,14 @@ from .relationships import (
     plot_pairwise_relationship_grid,
     plot_scatter_pair,
 )
+from .basic_series import (
+    PlotChoice,
+    PlotStyle,
+    plot_basic_series,
+    plot_dual_time_series,
+    recommended_style,
+    resample_series_for_plot,
+)
 from .seasonal_profiles import (
     plot_daylight_hours,
     plot_diurnal_cycle,
@@ -54,6 +62,12 @@ __all__ = [
     "plot_hexbin_with_third_variable",
     "plot_pairwise_relationship_grid",
     "plot_scatter_pair",
+    "PlotChoice",
+    "PlotStyle",
+    "plot_basic_series",
+    "plot_dual_time_series",
+    "recommended_style",
+    "resample_series_for_plot",
     "plot_daylight_hours",
     "plot_diurnal_cycle",
     "plot_seasonal_hourly_profiles",

meteo/plots/basic_series.py

@@ -0,0 +1,350 @@
+"""Tracés simples et réutilisables pour les séries temporelles de base."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+from enum import Enum
+from pathlib import Path
+from typing import Callable
+
+import matplotlib.dates as mdates
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+
+from meteo.dataset import _circular_mean_deg
+from meteo.variables import Variable
+
+from .base import export_plot_dataset
+
+__all__ = [
+    "PlotStyle",
+    "PlotChoice",
+    "recommended_style",
+    "resample_series_for_plot",
+    "plot_basic_series",
+    "plot_dual_time_series",
+]
+
+
+class PlotStyle(str, Enum):
+    LINE = "line"
+    AREA = "area"
+    BAR = "bar"
+    SCATTER = "scatter"
+
+
+@dataclass(frozen=True)
+class PlotChoice:
+    """Configuration par variable : style et fonction d'agrégation."""
+
+    style: PlotStyle
+    agg: Callable[[pd.Series], float] | str = "mean"
+
+
+DEFAULT_CHOICES: dict[str, PlotChoice] = {
+    # Variations continues : lignes ou aires.
+    "temperature": PlotChoice(PlotStyle.LINE, "mean"),
+    "pressure": PlotChoice(PlotStyle.LINE, "mean"),
+    "humidity": PlotChoice(PlotStyle.AREA, "mean"),
+    "illuminance": PlotChoice(PlotStyle.AREA, "mean"),
+    "sun_elevation": PlotChoice(PlotStyle.AREA, "mean"),
+    # Variables dont la perception bénéficie d'autres représentations.
+    "rain_rate": PlotChoice(PlotStyle.BAR, "mean"),
+    "wind_speed": PlotChoice(PlotStyle.LINE, "mean"),
+    "wind_direction": PlotChoice(PlotStyle.SCATTER, _circular_mean_deg),
+}
+
+# Palette douce mais contrastée, associée aux variables.
+PALETTE = {
+    "temperature": "#d1495b",
+    "pressure": "#5c677d",
+    "humidity": "#2c7bb6",
+    "rain_rate": "#1b9e77",
+    "illuminance": "#f4a259",
+    "wind_speed": "#118ab2",
+    "wind_direction": "#8e6c8a",
+    "sun_elevation": "#f08c42",
+}
+DEFAULT_COLOR = "#386cb0"
+
+
+def recommended_style(variable: Variable, override: str | None = None) -> PlotChoice:
+    """Retourne le style/agrégation par défaut, ou une surcharge utilisateur."""
+
+    if override:
+        style = PlotStyle(override)
+        agg = DEFAULT_CHOICES.get(variable.key, PlotChoice(style)).agg
+        return PlotChoice(style, agg)
+    return DEFAULT_CHOICES.get(variable.key, PlotChoice(PlotStyle.LINE))
+
+
+def _nice_frequencies() -> list[tuple[str, pd.Timedelta]]:
+    return [
+        ("5min", pd.Timedelta(minutes=5)),
+        ("10min", pd.Timedelta(minutes=10)),
+        ("15min", pd.Timedelta(minutes=15)),
+        ("30min", pd.Timedelta(minutes=30)),
+        ("1h", pd.Timedelta(hours=1)),
+        ("3h", pd.Timedelta(hours=3)),
+        ("6h", pd.Timedelta(hours=6)),
+        ("12h", pd.Timedelta(hours=12)),
+        ("1d", pd.Timedelta(days=1)),
+        ("3d", pd.Timedelta(days=3)),
+        ("7d", pd.Timedelta(days=7)),
+    ]
+
+
+def _auto_resample_frequency(index: pd.DatetimeIndex, *, target_points: int = 420) -> str:
+    """Choisit une fréquence qui limite le nombre de points tout en conservant la forme générale."""
+
+    if index.empty or len(index) < 2:
+        return "1h"
+
+    span = index.max() - index.min()
+    if span <= pd.Timedelta(0):
+        return "1h"
+
+    for label, delta in _nice_frequencies():
+        if span / delta <= target_points:
+            return label
+
+    return _nice_frequencies()[-1][0]
+
+
+def _format_time_axis(ax: plt.Axes) -> None:
+    locator = mdates.AutoDateLocator(minticks=4, maxticks=8)
+    formatter = mdates.ConciseDateFormatter(locator, formats=["%Y", "%b", "%d", "%d %H:%M", "%H:%M", "%S"])
+    ax.xaxis.set_major_locator(locator)
+    ax.xaxis.set_major_formatter(formatter)
+
+
+def _infer_bar_width(index: pd.DatetimeIndex) -> float:
+    """
+    Calcule une largeur de barre raisonnable (en jours) pour les histogrammes temporels.
+    """
+
+    if len(index) < 2:
+        return 0.3  # ~7 heures, pour rendre le point visible même isolé
+
+    diffs = np.diff(index.asi8)  # nanosecondes
+    median_ns = float(np.median(diffs))
+    if not np.isfinite(median_ns) or median_ns <= 0:
+        return 0.1
+    return pd.to_timedelta(median_ns, unit="ns") / pd.Timedelta(days=1) * 0.8
+
+
+def _ensure_datetime_index(series: pd.Series) -> pd.Series:
+    if not isinstance(series.index, pd.DatetimeIndex):
+        raise TypeError("Une série temporelle (DatetimeIndex) est attendue pour le tracé.")
+    return series
+
+
+def _series_color(variable: Variable) -> str:
+    if variable.key in PALETTE:
+        return PALETTE[variable.key]
+    return PALETTE.get(variable.column, DEFAULT_COLOR)
+
+
+def _format_label(var: Variable) -> str:
+    unit_text = f" ({var.unit})" if var.unit else ""
+    return f"{var.label}{unit_text}"
+
+
+def resample_series_for_plot(
+    series: pd.Series,
+    *,
+    variable: Variable,
+    freq: str | None = None,
+    target_points: int = 420,
+) -> tuple[pd.Series, str]:
+    """
+    Prépare une série pour l'affichage : resample et agrégation adaptés à la variable.
+    """
+
+    _ensure_datetime_index(series)
+
+    if freq is None:
+        freq = _auto_resample_frequency(series.index, target_points=target_points)
+
+    agg_func = DEFAULT_CHOICES.get(variable.key, PlotChoice(PlotStyle.LINE)).agg
+    resampled = series.resample(freq).agg(agg_func).dropna()
+    return resampled, freq
+
+
+def plot_basic_series(
+    series: pd.Series,
+    *,
+    variable: Variable,
+    output_path: str | Path,
+    style: PlotStyle,
+    title: str,
+    ylabel: str,
+    annotate_freq: str | None = None,
+) -> Path:
+    """
+    Trace une série temporelle avec un style simple (ligne, aire, barres, nuage de points).
+    """
+
+    _ensure_datetime_index(series)
+
+    if series.empty:
+        raise ValueError(f"Aucune donnée disponible pour {variable.key} après filtrage.")
+
+    output_path = Path(output_path)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    color = _series_color(variable)
+    x = mdates.date2num(series.index)
+    values = series.to_numpy(dtype=float)
+
+    fig, ax = plt.subplots(figsize=(11, 4.2))
+    if style is PlotStyle.LINE:
+        ax.plot_date(x, values, "-", linewidth=1.8, color=color, label=variable.label)
+    elif style is PlotStyle.AREA:
+        ax.fill_between(x, values, step="mid", color=color, alpha=0.2)
+        ax.plot_date(x, values, "-", linewidth=1.6, color=color)
+    elif style is PlotStyle.BAR:
+        width = _infer_bar_width(series.index)
+        ax.bar(x, values, width=width, color=color, edgecolor=color, linewidth=0.5, alpha=0.85)
+    elif style is PlotStyle.SCATTER:
+        ax.scatter(x, values, s=16, color=color, alpha=0.9)
+    else:
+        raise ValueError(f"Style inconnu : {style}")
+
+    ax.set_title(title)
+    ax.set_ylabel(ylabel)
+    _format_time_axis(ax)
+    ax.grid(True, color="#e0e0e0", linewidth=0.8, alpha=0.7)
+    ax.margins(x=0.02, y=0.05)
+
+    if annotate_freq:
+        ax.text(
+            0.99,
+            0.02,
+            f"Agrégation : {annotate_freq}",
+            transform=ax.transAxes,
+            ha="right",
+            va="bottom",
+            fontsize=9,
+            color="#555555",
+        )
+
+    fig.tight_layout()
+    fig.savefig(output_path, dpi=150)
+    plt.close(fig)
+
+    export_plot_dataset(series.to_frame(name=variable.column), output_path)
+    return output_path.resolve()
+
+
+def _draw_series(ax: plt.Axes, series: pd.Series, *, choice: PlotChoice, color: str, label: str):
+    x = mdates.date2num(series.index)
+    values = series.to_numpy(dtype=float)
+
+    if choice.style is PlotStyle.LINE:
+        return ax.plot_date(x, values, "-", linewidth=1.8, color=color, label=label)
+    if choice.style is PlotStyle.AREA:
+        ax.fill_between(x, values, step="mid", color=color, alpha=0.15)
+        return ax.plot_date(x, values, "-", linewidth=1.6, color=color, label=label)
+    if choice.style is PlotStyle.BAR:
+        width = _infer_bar_width(series.index) * 0.9
+        return ax.bar(x, values, width=width, color=color, edgecolor=color, linewidth=0.5, alpha=0.75, label=label)
+    if choice.style is PlotStyle.SCATTER:
+        return ax.scatter(x, values, s=16, color=color, alpha=0.9, label=label)
+    raise ValueError(f"Style inconnu : {choice.style}")
+
+
+def plot_dual_time_series(
+    series_left: pd.Series,
+    variable_left: Variable,
+    choice_left: PlotChoice,
+    series_right: pd.Series,
+    variable_right: Variable,
+    choice_right: PlotChoice,
+    *,
+    output_path: str | Path,
+    title: str,
+    annotate_freq: str | None = None,
+) -> Path:
+    """Superpose deux séries temporelles (axes Y séparés) avec styles adaptés."""
+
+    _ensure_datetime_index(series_left)
+    _ensure_datetime_index(series_right)
+
+    if series_left.empty or series_right.empty:
+        raise ValueError("Les séries à tracer ne peuvent pas être vides.")
+
+    output_path = Path(output_path)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    color_left = _series_color(variable_left)
+    color_right = _series_color(variable_right)
+
+    fig, ax_left = plt.subplots(figsize=(11, 4.6))
+    ax_right = ax_left.twinx()
+
+    artists_left = _draw_series(
+        ax_left,
+        series_left,
+        choice=choice_left,
+        color=color_left,
+        label=_format_label(variable_left),
+    )
+    artists_right = _draw_series(
+        ax_right,
+        series_right,
+        choice=choice_right,
+        color=color_right,
+        label=_format_label(variable_right),
+    )
+
+    ax_left.set_ylabel(_format_label(variable_left), color=color_left)
+    ax_right.set_ylabel(_format_label(variable_right), color=color_right)
+    ax_left.tick_params(axis="y", labelcolor=color_left)
+    ax_right.tick_params(axis="y", labelcolor=color_right)
+
+    _format_time_axis(ax_left)
+    ax_left.grid(True, color="#e0e0e0", linewidth=0.8, alpha=0.7)
+    ax_left.margins(x=0.02, y=0.05)
+    ax_right.margins(x=0.02, y=0.05)
+    ax_left.set_title(title)
+
+    handles = []
+    labels = []
+    for artist in artists_left if isinstance(artists_left, list) else [artists_left]:
+        handles.append(artist)
+        labels.append(artist.get_label())
+    if isinstance(artists_right, list):
+        handles.extend(artists_right)
+        labels.extend([a.get_label() for a in artists_right])
+    else:
+        handles.append(artists_right)
+        labels.append(artists_right.get_label())
+
+    ax_left.legend(handles, labels, loc="upper left")
+
+    if annotate_freq:
+        ax_left.text(
+            0.99,
+            0.02,
+            f"Agrégation : {annotate_freq}",
+            transform=ax_left.transAxes,
+            ha="right",
+            va="bottom",
+            fontsize=9,
+            color="#555555",
+        )
+
+    fig.tight_layout()
+    fig.savefig(output_path, dpi=150)
+    plt.close(fig)
+
+    export_plot_dataset(
+        pd.concat(
+            {variable_left.column: series_left, variable_right.column: series_right},
+            axis=1,
+        ),
+        output_path,
+    )
+    return output_path.resolve()