Premiers éléments de l'étude

lib/plots/parts_per_set.py

@@ -0,0 +1,110 @@
+"""Graphiques sur la taille moyenne des sets (pièces par set)."""
+
+from pathlib import Path
+from typing import Dict, Iterable, List, Tuple
+
+import matplotlib.pyplot as plt
+
+from lib.filesystem import ensure_parent_dir
+from lib.milestones import load_milestones
+from lib.rebrickable.stats import read_rows
+
+
+def compute_average_parts_per_set(rows: Iterable[dict]) -> List[Tuple[int, float]]:
+    """Calcule la moyenne annuelle de pièces par set."""
+    per_year: Dict[int, Dict[str, int]] = {}
+    for row in rows:
+        year = int(row["year"])
+        per_year[year] = per_year.get(year, {"parts": 0, "sets": 0})
+        per_year[year]["parts"] += int(row["num_parts"])
+        per_year[year]["sets"] += 1
+    results: List[Tuple[int, float]] = []
+    for year in sorted(per_year):
+        totals = per_year[year]
+        results.append((year, totals["parts"] / totals["sets"]))
+    return results
+
+
+def compute_rolling_mean(series: List[Tuple[int, float]], window: int) -> List[Tuple[int, float]]:
+    """Calcule la moyenne glissante sur une fenêtre donnée."""
+    values = [value for _, value in series]
+    years = [year for year, _ in series]
+    rolling: List[Tuple[int, float]] = []
+    for index in range(len(values)):
+        if index + 1 < window:
+            rolling.append((years[index], 0.0))
+        else:
+            window_values = values[index - window + 1 : index + 1]
+            rolling.append((years[index], sum(window_values) / window))
+    return rolling
+
+
+def plot_parts_per_set(
+    enriched_sets_path: Path,
+    milestones_path: Path,
+    destination_path: Path,
+    rolling_window: int = 3,
+) -> None:
+    """Génère un graphique de la moyenne annuelle et glissante des pièces par set."""
+    sets_rows = read_rows(enriched_sets_path)
+    milestones = load_milestones(milestones_path)
+    annual_series = compute_average_parts_per_set(sets_rows)
+    rolling_series = compute_rolling_mean(annual_series, rolling_window)
+    years = [year for year, _ in annual_series]
+    annual_values = [value for _, value in annual_series]
+    rolling_values = [value for _, value in rolling_series]
+
+    fig, ax = plt.subplots(figsize=(12, 6))
+    ax.plot(years, annual_values, marker="o", color="#2ca02c", label="Moyenne annuelle (pièces/set)")
+    ax.plot(
+        years,
+        rolling_values,
+        marker="^",
+        color="#9467bd",
+        label=f"Moyenne glissante {rolling_window} ans (pièces/set)",
+    )
+    ax.set_xlabel("Année")
+    ax.set_ylabel("Pièces par set")
+    ax.set_title("Évolution de la taille moyenne des sets (thèmes filtrés)")
+    ax.grid(True, linestyle="--", alpha=0.3)
+    ax.set_xlim(min(years) - 0.4, max(years) + 0.4)
+    ax.set_xticks(list(range(min(years), max(years) + 1)))
+    ax.tick_params(axis="x", labelrotation=45)
+
+    peak = max(max(annual_values), max(rolling_values))
+    top_limit = peak * 2
+    milestones_in_range = sorted(
+        [m for m in milestones if min(years) <= m["year"] <= max(years)],
+        key=lambda m: (m["year"], m["description"]),
+    )
+    milestone_offsets: Dict[int, int] = {}
+    offset_step = 0.4
+    max_offset = 0
+    for milestone in milestones_in_range:
+        year = milestone["year"]
+        count_for_year = milestone_offsets.get(year, 0)
+        milestone_offsets[year] = count_for_year + 1
+        horizontal_offset = offset_step * (count_for_year // 2 + 1)
+        max_offset = max(max_offset, count_for_year)
+        if count_for_year % 2 == 1:
+            horizontal_offset *= -1
+        text_x = year + horizontal_offset
+        ax.axvline(year, color="#d62728", linestyle="--", linewidth=1, alpha=0.65)
+        ax.text(
+            text_x,
+            top_limit,
+            milestone["description"],
+            rotation=90,
+            verticalalignment="top",
+            horizontalalignment="center",
+            fontsize=8,
+            color="#d62728",
+        )
+
+    ax.set_ylim(0, top_limit * (1 + max_offset * 0.02))
+    ax.legend(loc="upper left", bbox_to_anchor=(1.12, 1))
+
+    ensure_parent_dir(destination_path)
+    fig.tight_layout()
+    fig.savefig(destination_path, dpi=150)
+    plt.close(fig)