latex

task1/08_sensitivity.py

@@ -44,7 +44,8 @@ def truncation_correction(mu, sigma, C, p_thresh):
     if p_trunc < p_thresh:
         return mu, p_trunc, False
     else:
-        mu_tilde = mu * (1 + 0.4 * p_trunc)
+        # 与主流程 `02_demand_correction.py` 保持一致：线性修正系数 0.1
+        mu_tilde = mu * (1 + 0.1 * p_trunc)
         return mu_tilde, p_trunc, True
 
 

task1/export_paper_tables.py

@@ -0,0 +1,121 @@
+from __future__ import annotations
+
+import math
+from pathlib import Path
+
+import pandas as pd
+
+
+ROOT = Path(__file__).resolve().parent
+OUT_DIR = ROOT / "paper_tables"
+
+
+def _to_int_series(s: pd.Series) -> pd.Series:
+    return s.map(lambda x: int(round(float(x))) if pd.notna(x) else x)
+
+
+def _to_float_series(s: pd.Series, ndigits: int) -> pd.Series:
+    return s.map(lambda x: round(float(x), ndigits) if pd.notna(x) else x)
+
+
+def _format_table1_metrics_summary(df: pd.DataFrame) -> pd.DataFrame:
+    out = df.copy()
+    out["E1_total_service"] = _to_int_series(out["E1_total_service"])
+    out["E2_quality_weighted"] = _to_int_series(out["E2_quality_weighted"])
+    out["F1_gini"] = _to_float_series(out["F1_gini"], 4)
+    out["F2_min_satisfaction"] = _to_float_series(out["F2_min_satisfaction"], 2)
+    out["F3_cv_satisfaction"] = _to_float_series(out["F3_cv_satisfaction"], 4)
+    out["E1_total_service_pct"] = _to_float_series(out["E1_total_service_pct"], 2)
+    out["E2_quality_weighted_pct"] = _to_float_series(out["E2_quality_weighted_pct"], 2)
+    return out
+
+
+def _format_table2_corrected_sites(df: pd.DataFrame) -> pd.DataFrame:
+    out = df.copy()
+    out["site_id"] = _to_int_series(out["site_id"])
+    out["mu"] = _to_float_series(out["mu"], 1)
+    out["sigma"] = _to_float_series(out["sigma"], 1)
+    out["p_trunc"] = _to_float_series(out["p_trunc"], 3)
+    out["mu_tilde"] = _to_float_series(out["mu_tilde"], 1)
+    out["mu_delta"] = _to_float_series(out["mu_delta"], 1)
+    out["mu_delta_pct"] = _to_float_series(out["mu_delta_pct"], 1)
+    return out
+
+
+def _format_table3_backtest_summary(df: pd.DataFrame) -> pd.DataFrame:
+    out = df.copy()
+    out["value"] = out["value"].map(lambda x: round(float(x), 6) if pd.notna(x) else x)
+    out["p_value"] = out["p_value"].map(lambda x: round(float(x), 6) if pd.notna(x) else x)
+    return out
+
+
+def _format_appendix_site_details(df: pd.DataFrame) -> pd.DataFrame:
+    out = df.copy()
+    out["site_id"] = _to_int_series(out["site_id"])
+    out["mu"] = _to_float_series(out["mu"], 1)
+    out["mu_tilde"] = _to_float_series(out["mu_tilde"], 1)
+    out["k"] = _to_int_series(out["k"])
+    out["annual_service"] = _to_int_series(out["annual_service"])
+    out["r"] = _to_float_series(out["r"], 4)
+    out["quality_factor"] = _to_float_series(out["quality_factor"], 4)
+    out["service_quality_weighted"] = _to_int_series(out["service_quality_weighted"])
+    return out
+
+
+def export_tables() -> list[Path]:
+    OUT_DIR.mkdir(parents=True, exist_ok=True)
+    written: list[Path] = []
+
+    # Table 1: overall performance comparison
+    metrics_xlsx = ROOT / "04_metrics.xlsx"
+    metrics = pd.read_excel(metrics_xlsx, sheet_name="metrics_summary")
+    metrics_paper = _format_table1_metrics_summary(metrics)
+    p1 = OUT_DIR / "table1_metrics_summary.csv"
+    metrics_paper.to_csv(p1, index=False, encoding="utf-8-sig")
+    written.append(p1)
+
+    # Table 2: truncation correction (corrected sites)
+    backtest_xlsx = ROOT / "07_backtest.xlsx"
+    corrected = pd.read_excel(backtest_xlsx, sheet_name="corrected_sites")
+    corrected = corrected.copy()
+    corrected["mu_delta"] = corrected["mu_tilde"] - corrected["mu"]
+    corrected["mu_delta_pct"] = corrected["mu_delta"] / corrected["mu"] * 100
+    corrected_paper = _format_table2_corrected_sites(corrected)
+    p2 = OUT_DIR / "table2_corrected_sites.csv"
+    corrected_paper.to_csv(p2, index=False, encoding="utf-8-sig")
+    written.append(p2)
+
+    # Table 3: backtest & fit statistics
+    summary = pd.read_excel(backtest_xlsx, sheet_name="summary_metrics")
+    summary_paper = _format_table3_backtest_summary(summary)
+    p3 = OUT_DIR / "table3_backtest_summary.csv"
+    summary_paper.to_csv(p3, index=False, encoding="utf-8-sig")
+    written.append(p3)
+
+    # Table 4: constraint validation results
+    validate_xlsx = ROOT / "06_validate.xlsx"
+    validation = pd.read_excel(validate_xlsx, sheet_name="validation_results")
+    p4 = OUT_DIR / "table4_validation_results.csv"
+    validation.to_csv(p4, index=False, encoding="utf-8-sig")
+    written.append(p4)
+
+    # Appendix: per-site allocation details (optional for paper appendix)
+    site_details = pd.read_excel(metrics_xlsx, sheet_name="site_details")
+    site_details_paper = _format_appendix_site_details(site_details)
+    p5 = OUT_DIR / "appendix_site_details.csv"
+    site_details_paper.to_csv(p5, index=False, encoding="utf-8-sig")
+    written.append(p5)
+
+    return written
+
+
+def main() -> None:
+    written = export_tables()
+    rel = [p.relative_to(ROOT) for p in written]
+    print("Wrote:")
+    for p in rel:
+        print(f" - {p}")
+
+
+if __name__ == "__main__":
+    main()

task1/paper_tables/appendix_site_details.csv

@@ -0,0 +1,71 @@
+site_id,site_name,mu,mu_tilde,k,annual_service,r,quality_factor,service_quality_weighted
+1.0,MFP American Legion - Binghamton,200,200,14.0,2803,14.0,1.0,2803
+2.0,MFP Avoca,315,323,22.0,6921,21.4249,0.7947,5500
+3.0,MFP Bath,279,279,19.0,5310,19.0,0.8946,4750
+4.0,MFP Beaver Dams,171,171,12.0,2048,12.0,1.0,2048
+5.0,MFP Birnie Transportation Services,213,213,15.0,3201,15.0,1.0,3201
+6.0,MFP Boys and Girls Club,211,211,15.0,3162,15.0,1.0,3162
+7.0,MFP Bradford,122,122,9.0,1100,9.0,1.0,1100
+8.0,MFP Campbell,168,168,12.0,2022,12.0,1.0,2022
+9.0,MFP Canisteo,177,177,13.0,2301,13.0,1.0,2301
+10.0,MFP Colesville,197,197,14.0,2763,14.0,1.0,2763
+11.0,MFP College Corning Community College,251,251,18.0,4518,18.0,0.996,4500
+12.0,MFP College Ithaca College,138,138,10.0,1383,10.0,1.0,1383
+13.0,MFP College TC3 -College,262,266,19.0,4968,18.686,0.956,4750
+14.0,MFP Conklin- Maines Community Center,153,153,11.0,1685,11.0,1.0,1685
+15.0,MFP Danby,160,160,12.0,1923,12.0,1.0,1923
+16.0,MFP Deposit,157,157,11.0,1722,11.0,1.0,1722
+17.0,MFP Endwell United Methodist Church,285,289,20.0,5705,19.7173,0.8764,5000
+18.0,MFP Erin,174,174,13.0,2261,13.0,1.0,2261
+19.0,MFP First Assembly Of God Church,146,146,11.0,1606,11.0,1.0,1606
+20.0,MFP Lamphear Court,126,126,9.0,1134,9.0,1.0,1134
+21.0,MFP Lansing,181,181,13.0,2353,13.0,1.0,2353
+22.0,MFP Lindley,233,233,16.0,3726,16.0,1.0,3726
+23.0,MFP Millport,166,166,12.0,1992,12.0,1.0,1992
+24.0,MFP Montour Falls-Schuyler County Human Services Complex,149,149,11.0,1643,11.0,1.0,1643
+25.0,MFP Nichols-The Creamery,122,122,9.0,1102,9.0,1.0,1102
+26.0,MFP Owego VFW,176,176,13.0,2291,13.0,1.0,2291
+27.0,MFP Prattsburgh,144,144,11.0,1590,11.0,1.0,1590
+28.0,MFP Rathbone,269,269,19.0,5113,19.0,0.9291,4750
+29.0,MFP Reach for Christ Church Freeville,220,220,16.0,3520,16.0,1.0,3520
+30.0,MFP Redeemer Lutheran Church,231,233,16.0,3690,15.8404,1.0,3690
+31.0,MFP Rehoboth Deliverance Ministry,236,236,17.0,4010,17.0,1.0,4010
+32.0,MFP Richford,266,266,19.0,5052,19.0,0.9402,4750
+33.0,MFP Saint Mary Recreation Center,148,148,11.0,1631,11.0,1.0,1631
+34.0,MFP Salvation Army Ithaca,181,181,13.0,2355,13.0,1.0,2355
+35.0,MFP Schuyler Outreach,139,139,10.0,1389,10.0,1.0,1389
+36.0,MFP Senior - Addison Place Apartments,30,30,3.0,90,3.0,1.0,90
+37.0,MFP Senior - Bragg,67,67,5.0,334,5.0,1.0,334
+38.0,MFP Senior - Carpenter Apartments,31,31,3.0,93,3.0,1.0,93
+39.0,MFP Senior - Cayuga Meadows,26,26,3.0,78,3.0,1.0,78
+40.0,MFP Senior - CFS Lakeview,112,112,8.0,896,8.0,1.0,896
+41.0,MFP Senior - Conifer Village,34,34,3.0,101,3.0,1.0,101
+42.0,MFP Senior - Corning Senior Center,75,75,6.0,450,6.0,1.0,450
+43.0,MFP Senior - Dayspring,77,77,6.0,463,6.0,1.0,463
+44.0,MFP Senior - East Hill Senior Living,40,40,4.0,159,4.0,1.0,159
+45.0,"MFP Senior - Elizabeth Square, Waverly",29,29,3.0,87,3.0,1.0,87
+46.0,MFP Senior - Ellis Hollow,25,25,3.0,74,3.0,1.0,74
+47.0,MFP Senior - Flannery,62,62,5.0,309,5.0,1.0,309
+48.0,MFP Senior - Harry L Apartments,33,33,3.0,99,3.0,1.0,99
+49.0,MFP Senior - Jefferson Village,25,25,3.0,74,3.0,1.0,74
+50.0,MFP Senior - Lincoln Court,26,26,3.0,78,3.0,1.0,78
+51.0,MFP Senior - Long Meadow Senior Housing,35,35,3.0,104,3.0,1.0,104
+52.0,MFP Senior - Metro Plaza Apartments,56,56,5.0,282,5.0,1.0,282
+53.0,MFP Senior - North Shore Towers,58,58,5.0,292,5.0,1.0,292
+54.0,"MFP Senior - Northern Broome Senior Center, Whitney Point",51,51,4.0,204,4.0,1.0,204
+55.0,MFP Senior - Park Terrace Congregate Apartments,24,24,3.0,73,3.0,1.0,73
+56.0,MFP Senior - Springview Apartments,28,28,3.0,83,3.0,1.0,83
+57.0,MFP Senior - Titus Towers,73,73,6.0,437,6.0,1.0,437
+58.0,MFP Senior - Villa Serene,70,70,6.0,418,6.0,1.0,418
+59.0,MFP Senior - Village Square/Manor,34,34,3.0,103,3.0,1.0,103
+60.0,MFP Senior - Wells Apartments,24,24,3.0,70,3.0,1.0,70
+61.0,MFP Senior - Woodsedge Apartments,17,17,2.0,34,2.0,1.0,34
+62.0,MFP The Love Church,259,259,18.0,4668,18.0,0.964,4500
+63.0,MFP Troupsburg,149,149,11.0,1636,11.0,1.0,1636
+64.0,MFP Tuscarora,193,193,14.0,2697,14.0,1.0,2697
+65.0,MFP Van Etten,214,214,15.0,3206,15.0,1.0,3206
+66.0,MFP Waverly,397,429,29.0,11502,26.8129,0.6303,7250
+67.0,MFP Wayland,180,180,13.0,2345,13.0,1.0,2345
+68.0,MFP Whitney Point,203,203,14.0,2836,14.0,1.0,2836
+69.0,MFP Windsor,201,201,14.0,2813,14.0,1.0,2813
+70.0,MFP Woodhull,176,176,13.0,2288,13.0,1.0,2288

task1/paper_tables/table1_metrics_summary.csv

@@ -0,0 +1,5 @@
+method,E1_total_service,E2_quality_weighted,F1_gini,F2_min_satisfaction,F3_cv_satisfaction,E1_total_service_pct,E2_quality_weighted_pct
+Recommended (μ̃ proportional),139469,131462,0.3137,2.0,0.5603,0,0
+Baseline 1: Uniform,104797,101309,0.0244,9.2458,0.0486,-25,-23
+Baseline 2: 2019 Scaled,104071,100264,0.0915,5.0,0.1759,-25,-24
+Baseline 3: μ proportional (no correction),139129,131397,0.313,2.0,0.5561,0,0

task1/paper_tables/table2_corrected_sites.csv

@@ -0,0 +1,6 @@
+site_id,site_name,mu,sigma,p_trunc,mu_tilde,mu_delta,mu_delta_pct
+2.0,MFP Avoca,315,57.3497,0.2684,323,8,3.0
+13.0,MFP College TC3 -College,262,91.9929,0.168,266,4,2.0
+17.0,MFP Endwell United Methodist Church,285,60.7815,0.1434,289,4,1.0
+30.0,MFP Redeemer Lutheran Church,231,93.4787,0.1007,233,2,1.0
+66.0,MFP Waverly,397,51.8754,0.8157,429,32,8.0

task1/paper_tables/table3_backtest_summary.csv

@@ -0,0 +1,10 @@
+metric,value,p_value
+"corr(visits_2019, μ)",0.0349,0.774
+"corr(visits_2019, μ̃)",0.0364,0.7645
+sites_increased,36.0,
+sites_decreased,32.0,
+sites_unchanged,2.0,
+CV(k/μ̃),0.1974,
+service_2019_actual,102473.0,
+service_model,139469.0768,
+improvement_pct,34.6117,

task1/paper_tables/table4_validation_results.csv

@@ -0,0 +1,7 @@
+constraint,formula,expected,actual,passed,details
+C1: 资源约束,Σk_i = 730,730,730,True,总访问次数 = 730
+C2: 覆盖约束,k_i >= 1,>= 1,min = 2,True,"最小访问次数 = 2, 违反站点数 = 0"
+C3: 日容量约束,每日站点数 = 2,所有天 = 2,"min=2, max=2",True,不满足的天数 = 0
+C4: 无重复约束,site_1 ≠ site_2 (同一天),无重复,重复天数 = 0,True,无
+C5: 排程一致性,排程次数 = k_i,所有站点一致,不一致站点数 = 0,True,全部一致
+C6: 总天数,日历天数 = 365,365,365,True,日历包含 365 天