p1: fig

p1/sensitivity_analysis.py

@@ -585,6 +585,96 @@ def plot_sensitivity_analysis(
     return fig
 
 
+def plot_tornado_chart_standalone(
+    payload_df: pd.DataFrame,
+    elevator_df: pd.DataFrame, 
+    frequency_df: pd.DataFrame,
+    alpha_df: pd.DataFrame,
+    save_path: str = '/Volumes/Files/code/mm/20260130_b/p1/tornado_chart.png'
+):
+    """
+    单独绘制 Tornado Chart（敏感性汇总图）
+    - XY轴反转（垂直柱状图）
+    - 长宽比 0.618
+    - 无标题
+    - 有图例
+    - 低饱和度色系
+    """
+    # 使用0.618黄金比例
+    fig_width = 8
+    fig_height = fig_width * 0.618
+    fig, ax = plt.subplots(figsize=(fig_width, fig_height))
+    
+    # 计算敏感性指标
+    baseline_T = 139
+    
+    sensitivities = [
+        ('Payload\n(100→150t)', 
+         (payload_df['optimal_T_lambda504'].iloc[0] - baseline_T) / baseline_T * 100,
+         (payload_df['optimal_T_lambda504'].iloc[-1] - baseline_T) / baseline_T * 100),
+        ('Elevator Cap.\n(80%→120%)', 
+         (elevator_df['optimal_T_lambda504'].iloc[0] - baseline_T) / baseline_T * 100,
+         (elevator_df['optimal_T_lambda504'].iloc[-1] - baseline_T) / baseline_T * 100),
+        ('Launch Freq.\n(0.5→2/day)', 
+         (frequency_df['optimal_T_lambda504'].iloc[0] - baseline_T) / baseline_T * 100,
+         (frequency_df['optimal_T_lambda504'].iloc[-1] - baseline_T) / baseline_T * 100),
+        ('Struct. Coef.\n(0.06→0.14)', 
+         (alpha_df['optimal_T_lambda504'].iloc[0] - baseline_T) / baseline_T * 100,
+         (alpha_df['optimal_T_lambda504'].iloc[-1] - baseline_T) / baseline_T * 100),
+    ]
+    
+    # 低饱和度配色
+    color_positive = '#7A9CBF'  # 灰蓝色（参数增加导致T*增加）
+    color_negative = '#C4816B'  # 灰橙红色（参数增加导致T*减少）
+    
+    x_pos = np.arange(len(sensitivities))
+    bar_width = 0.6
+    
+    # 绘制柱状图（XY反转，使用垂直柱状图）
+    for i, (name, low, high) in enumerate(sensitivities):
+        bar_color = color_positive if high > low else color_negative
+        bar_height = high - low
+        bar_bottom = min(low, high)
+        
+        ax.bar(i, bar_height, bottom=bar_bottom, width=bar_width, 
+               color=bar_color, alpha=0.8, edgecolor='#555555', linewidth=0.8)
+        
+        # 标记端点
+        ax.plot(i, low, 'o', color='#444444', markersize=7, zorder=5)
+        ax.plot(i, high, '^', color='#444444', markersize=7, zorder=5)
+        ax.plot([i, i], [low, high], '-', color='#444444', linewidth=1.5, zorder=4)
+    
+    # 基准线
+    ax.axhline(y=0, color='#333333', linestyle='-', linewidth=1.2)
+    
+    # 设置轴
+    ax.set_xticks(x_pos)
+    ax.set_xticklabels([s[0] for s in sensitivities], fontsize=10)
+    ax.set_ylabel('Change in Optimal Timeline T* (%)', fontsize=11)
+    ax.grid(True, alpha=0.3, axis='y')
+    
+    # 添加图例
+    from matplotlib.patches import Patch
+    from matplotlib.lines import Line2D
+    legend_elements = [
+        Patch(facecolor=color_negative, edgecolor='#555555', alpha=0.8, 
+              label='Parameter ↑ → T* ↓'),
+        Patch(facecolor=color_positive, edgecolor='#555555', alpha=0.8, 
+              label='Parameter ↑ → T* ↑'),
+        Line2D([0], [0], marker='o', color='#444444', linestyle='None', 
+               markersize=6, label='Low value'),
+        Line2D([0], [0], marker='^', color='#444444', linestyle='None', 
+               markersize=6, label='High value'),
+    ]
+    ax.legend(handles=legend_elements, loc='upper right', fontsize=9, framealpha=0.9)
+    
+    plt.tight_layout()
+    plt.savefig(save_path, dpi=150, bbox_inches='tight')
+    print(f"Tornado chart saved to: {save_path}")
+    
+    return fig
+
+
 def plot_tradeoff_comparison(
     save_path: str = '/Volumes/Files/code/mm/20260130_b/p1/sensitivity_tradeoff.png'
 ):