36 lines
1.6 KiB
Markdown
36 lines
1.6 KiB
Markdown
# 20260116 美赛模拟：食物分发（MFP）
|
||
|
||
本仓库包含对 2019 年 MFP 站点数据的频次分配（任务一）与全年排班优化（任务二）脚本。
|
||
核心数据文件：`prob/MFP Regular Sites 2019.xlsx`。
|
||
|
||
## Task 1: 频次分配（Visit Frequency Allocation）
|
||
|
||
目标：在总车次约束下，为每个站点分配年度访问次数 `f_i`，并评估有效性/公平性（含最低10%平均、基尼系数等）。
|
||
|
||
- 运行：`python3 kmin_effectiveness.py`
|
||
- 输出（写入 `data/`）：
|
||
- `data/kmin_effectiveness.png`：k_min 与指标曲线图
|
||
- `data/kmin_effectiveness_data.csv`：每个 `k_min` 的汇总指标 + 各站点 `visits_01..visits_N`
|
||
- `data/kmin_effectiveness_sites.csv`：`visits_XX` 与站点名称/顺序映射
|
||
|
||
说明：`kmin_effectiveness.py` 当前使用 Monte Carlo（引入 `StDev(Demand per Visit)`）对有效性做多次模拟平均。
|
||
|
||
|
||
|
||
## Scheduling (Step 2)
|
||
|
||
Optimize a 365-day schedule with at most 2 visits per day and minimum gap constraints:
|
||
|
||
- `python3 scheduling_optimization.py --days 365 --daily-capacity 2 --gap-min 14`
|
||
- Outputs are written to `data/` (e.g., `data/schedule_optimized_kmin6.8_gap14.csv`), using `data/kmin_effectiveness_data.csv` as the frequency source.
|
||
|
||
### Visualization (Plan A)
|
||
|
||
- `python3 visualize_schedule.py`
|
||
- Outputs: `data/schedule_barcode_*.png` and `data/schedule_gap_deviation_*.png`
|
||
- Site label rule: remove first 4 chars, then take 12 chars.
|
||
|
||
|
||
|
||
|