Graphiques de corrélations binaires simples

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.
-Chaque figure déposée dans `figures/` possède son CSV compagnon exporté dans le dossier `data/` au même emplacement.
+## Superpositions simples
+
+```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

@@ -26,6 +26,7 @@ from .basic_series import (
     PlotChoice,
     PlotStyle,
     plot_basic_series,
+    plot_dual_time_series,
     recommended_style,
     resample_series_for_plot,
 )
@@ -64,6 +65,7 @@ __all__ = [
     "PlotChoice",
     "PlotStyle",
     "plot_basic_series",
+    "plot_dual_time_series",
     "recommended_style",
     "resample_series_for_plot",
     "plot_daylight_hours",

meteo/plots/basic_series.py

@@ -23,6 +23,7 @@ __all__ = [
     "recommended_style",
     "resample_series_for_plot",
     "plot_basic_series",
+    "plot_dual_time_series",
 ]
 
 
@@ -145,6 +146,11 @@ def _series_color(variable: Variable) -> str:
     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,
     *,
@@ -230,3 +236,115 @@ def plot_basic_series(
 
     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()