split,arm,n,mean_abs_log_error,mdape,p95_abs_log_error,bias_log,severe_rate,over_rate,coverage_50,coverage_90,mean_log_width_90 test,D (full partition),96935.0,0.1714733055793555,0.1054069371179375,0.622153490812197,0.0097700453442457,0.2180739670913498,0.1104038788879145,0.4463815959147882,0.8964048073451282,0.8299789424600684 test,D (paired subsample),80.0,0.16259817018566577,0.12179550958537777,0.5064464109033365,-0.030532750097691585,0.2125,0.075,0.425,0.875,0.8461984795362621 test,E,80.0,0.1655996928706865,0.1068212115441212,0.5075374999263345,0.0128334984990688,0.2375,0.125,0.4625,0.8875,0.5620513641013618 test,F,80.0,0.1587102857198576,0.0905148058205572,0.4990352168129302,0.0033337688954587,0.2125,0.0875,0.5125,0.8875,0.589534685295042 test,G2,80.0,0.1575785832187693,0.0865224989078201,0.4930899944960976,-0.0035708238551207,0.225,0.1,0.475,0.8875,0.6022991979979221 holdout 2026,D (full partition),83489.0,0.1799047200604636,0.1124254594153789,0.6433706067447303,0.0023498750222177,0.2393249410101929,0.1148294984968079,0.4343566218304208,0.8776365748781276,0.8270870792558116 holdout 2026,D (paired subsample),60.0,0.18889238482898604,0.11329143978728923,0.6893172438228633,0.05469563788313193,0.25,0.18333333333333332,0.36666666666666664,0.9,0.815463319740074 holdout 2026,E,60.0,0.1952999223487625,0.0860627025133548,0.8464376869734441,0.1103947605762593,0.2666666666666666,0.1833333333333333,0.4666666666666667,0.8,0.4968847958684848 holdout 2026,F,60.0,0.1858705315430335,0.0898781580032151,0.8051454444335208,0.1066006315050173,0.2333333333333333,0.1833333333333333,0.5166666666666667,0.8333333333333334,0.5077470175934654 holdout 2026,G2,60.0,0.1888295841931927,0.0888144439722939,0.82772064059921,0.1023344908826574,0.25,0.1833333333333333,0.5333333333333333,0.8333333333333334,0.5277165599452268 prospective,D (full partition),1182.0,0.2024250307340938,0.1294503431770919,0.7395463082579299,0.0387183286597189,0.2808798646362098,0.1565143824027072,0.4103214890016921,0.8663282571912013,0.8893791351125315 prospective,D (paired subsample),40.0,0.19746645075067892,0.09609054499971137,0.697952444673532,0.058937101469432915,0.275,0.15,0.35,0.9,0.9019311277535866 prospective,E,40.0,0.1865075397015597,0.0850668274545277,0.6448924880894997,0.058775830031803,0.3,0.175,0.425,0.75,0.5579650700737118 prospective,F,40.0,0.1787261001431398,0.0878879308262499,0.6600587868492497,0.0608762763662705,0.275,0.175,0.5,0.8,0.5741003394033062 prospective,G2,40.0,0.1873953939582804,0.096109380183864,0.6619796463871616,0.0577016364165637,0.3,0.175,0.45,0.825,0.5707408543553014