comparison,n,median_model,model_ci_lo,model_ci_hi,median_base,base_ci_lo,base_ci_hi,median_paired_diff,diff_ci_lo,diff_ci_hi,hodges_lehmann,hl_ci_lo,hl_ci_hi,wins,losses,rank_biserial,p_raw,direction,naive_direction,direction_conflict,family,p_holm,significant_raw,significant_holm,survives_holm,lost_to_holm
Random Forest,191,1.0548689753642493,1.043623390333624,1.0719095099288498,1.065386301706189,1.0523197553767492,1.0811572778619036,-0.016529854825019452,-0.02901232565876133,-0.009468101219284542,-0.02542647085162386,-0.04222171668103618,-0.011754568489765756,121,70,-0.31271815008726006,0.00017845241354529765,model better,model better,False,"ARCHER2 (fixed size, LOAO) vs constant baseline, intersection features",0.0003569048270905953,True,True,True,False
GBM quantile(0.5),191,1.0547408908806901,1.0468850902546476,1.0694610015846067,1.065386301706189,1.051871935581605,1.0811572778619036,-0.022823601662282034,-0.030466979751038403,-0.0072424011508025465,-0.02689380540639036,-0.04253169417676521,-0.01301828921349516,118,73,-0.3212260034904014,0.00011827393494640801,model better,model better,False,"ARCHER2 (fixed size, LOAO) vs constant baseline, intersection features",0.00035482180483922403,True,True,True,False
"MLP (32,16) alpha=1",191,1.0759852010868358,1.064662571856869,1.095836819421201,1.065386301706189,1.051871935581605,1.0798950986243512,-0.014745659823035995,-0.024571897414148225,-0.008323156210138238,-0.01662834837919913,-0.028182004928341775,-0.005009061116758788,119,72,-0.225021815008726,0.00700231109799172,model better,model worse,True,"ARCHER2 (fixed size, LOAO) vs constant baseline, intersection features",0.00700231109799172,True,True,True,False
Random Forest,191,1.186313203520252,1.131127483931828,1.2358106374376345,1.1662804304642018,1.1542301316792505,1.1995614330287676,-0.05364453834735006,-0.0768389451201632,-0.014904311333874709,-0.06117635282801903,-0.15439804065511173,-0.01422884177693596,113,78,-0.20636998254799302,0.013390740816933583,model better,model worse,True,"Cirrus fixed-size (LOAO) vs constant baseline, intersection features",0.04017222245080075,True,True,True,False
GBM quantile(0.5),191,1.2146392283998386,1.1639489396003473,1.3399782002162972,1.1662804304642018,1.1542301316792505,1.1995614330287676,-0.04369571926246585,-0.06961037005416282,0.0033413739995091785,-0.05906569684928403,-0.14095267890583474,-0.007436334620603672,105,86,-0.18346422338568935,0.02789963885766309,model better,model worse,True,"Cirrus fixed-size (LOAO) vs constant baseline, intersection features",0.05579927771532618,True,False,False,True
"MLP (32,16) alpha=1",191,1.2891524810818775,1.1633804148242104,1.40012340594109,1.1662804304642018,1.1542197040869924,1.1995614330287676,-0.022648810614095627,-0.06668455264752637,0.035404549238973404,-0.035610346566707085,-0.09927987550451382,0.03186384497270238,98,93,-0.08802356020942409,0.29146975316625234,model better,model worse,True,"Cirrus fixed-size (LOAO) vs constant baseline, intersection features",0.29146975316625234,False,False,False,False
Random Forest,162,1.0602727465183022,1.0474189072755955,1.0790067889706407,1.0616564240576782,1.0564703044897374,1.0677336514364464,-0.011874043644914734,-0.016006953084537794,-0.0061778450031103604,-0.007379485545828457,-0.014763627346343667,0.003350984209977544,101,61,-0.1400439294099826,0.12209592190094908,model better,model better,False,"Cirrus weak-scaled (LOAO) vs constant baseline, intersection features",0.3662877657028472,False,False,False,False
GBM quantile(0.5),162,1.0564776401462233,1.0457125683630175,1.071705016477591,1.0616564240576782,1.0565014196866194,1.0687540264386595,-0.008880613820546945,-0.01824306766376138,0.0014214947068833528,-0.006052115081926268,-0.01783676326896965,0.006869672050566257,91,71,-0.08884344467166554,0.32669116643979546,model better,model better,False,"Cirrus weak-scaled (LOAO) vs constant baseline, intersection features",0.36683238984340183,False,False,False,False
"MLP (32,16) alpha=1",162,1.0806619179542838,1.0693575096005508,1.1044122039906639,1.0616564240576782,1.0561286516131758,1.0677336514364464,0.008443226555118533,-0.0015829809832648145,0.019717070730792097,0.011105155957274682,-0.0034500040737683035,0.02882894055001836,71,91,0.1205029160039385,0.18341619492170091,model worse,model worse,False,"Cirrus weak-scaled (LOAO) vs constant baseline, intersection features",0.36683238984340183,False,False,False,False
+ analytic cycles (1 counter),191,1.105176891094316,1.0741777958425827,1.1542202190634971,1.1039687273047365,1.0940873262478994,1.1212514386335926,-0.03368150924639579,-0.047432430859893016,-0.004510385085699342,-0.03745005945544816,-0.05990461275564167,-0.0164049996194488,118,73,-0.30650087260034903,0.00023952646289968172,model better,model worse,True,"Counter-group ablation on ARCHER2, RF, vs CONFIG-ONLY",0.00023952646289968172,True,True,True,False
+ instruction ratios (1 set),191,1.079537375478765,1.0644150019163303,1.1099778427442166,1.1039687273047365,1.0940873262478994,1.1212514386335926,-0.03684475810826049,-0.046010869376939656,-0.021969859563218952,-0.051379813943108754,-0.0809399203796526,-0.032249454435724606,131,60,-0.4899650959860384,4.308274614946403e-09,model better,model better,False,"Counter-group ablation on ARCHER2, RF, vs CONFIG-ONLY",8.616549229892806e-09,True,True,True,False
+ all counters (5 sets),191,1.049636689675054,1.0372078678740209,1.072558149012392,1.1039687273047365,1.092567823051074,1.1212514386335926,-0.05291948537265534,-0.06677717871853894,-0.04165116643903022,-0.06982018963643644,-0.09787580452214718,-0.05170431544787937,152,39,-0.6855366492146597,2.109419701785495e-16,model better,model better,False,"Counter-group ablation on ARCHER2, RF, vs CONFIG-ONLY",6.328259105356485e-16,True,True,True,False
config only (free),191,1.1039687273047365,1.0940493386679788,1.1212514386335926,1.065386301706189,1.0523197553767492,1.0811572778619036,0.016470225751809897,0.008846096995574149,0.021218336164258167,0.017429662087458775,0.007648283504405065,0.028205212518489505,64,127,0.26570680628272253,0.0014509629899963354,model worse,model worse,False,"Counter-group ablation on ARCHER2, RF, vs CONSTANT baseline",0.004352888969989007,True,True,True,False
+ analytic cycles (1 counter),191,1.105176891094316,1.0759721159547042,1.1542202190634971,1.065386301706189,1.051871935581605,1.0811572778619036,0.0022398815218869217,-0.02039522474942257,0.012238594557215654,0.0007196632216421861,-0.01760439133073399,0.01853083015264356,95,96,0.007198952879581152,0.9312481425439445,model worse,model worse,False,"Counter-group ablation on ARCHER2, RF, vs CONSTANT baseline",0.9312481425439445,False,False,False,False
+ instruction ratios (1 set),191,1.079537375478765,1.0644150019163303,1.1099778427442166,1.065386301706189,1.051871935581605,1.0798950986243512,-0.0037808057068271506,-0.017626329578449562,0.006572075923746867,-0.011902981885326958,-0.029307425344937774,0.0031663988190435417,101,90,-0.1255453752181501,0.1324330049541339,model better,model worse,True,"Counter-group ablation on ARCHER2, RF, vs CONSTANT baseline",0.2648660099082678,False,False,False,False
+ all counters (5 sets),191,1.049636689675054,1.0371381416283774,1.0741672678561072,1.065386301706189,1.051871935581605,1.0798950986243512,-0.018425725185437614,-0.02691295200396837,-0.014606499991576705,-0.0292059927285776,-0.04381114530679656,-0.015729826159406806,130,61,-0.3450043630017452,3.555052018439601e-05,model better,model better,False,"Counter-group ablation on ARCHER2, RF, vs CONSTANT baseline",0.00014220208073758404,True,True,True,False
ARCHER2 -> Cirrus-fixed: RF,191,1.1943394962256582,1.1866635438944906,1.206112342667806,1.2018158544916426,1.1962524166984017,1.2159619486918882,-0.024137869922616906,-0.03432346052540525,-0.01395828127636256,-0.035701866295073714,-0.11661981189579372,-0.018641102896115556,124,67,-0.31653577661431065,0.00014855820665187153,model better,model better,False,"Cross-platform transfer vs constant, intersection features",0.0009078363763794523,True,True,True,False
ARCHER2 -> Cirrus-fixed: GBQ,191,1.1956493977766782,1.1762419876749353,1.2102422724213227,1.2018158544916426,1.1969564527874303,1.2159619486918882,-0.029810536175024938,-0.038606006244701785,-0.017111291471061607,-0.04824408362799354,-0.14440830888545572,-0.024347140696612802,122,69,-0.3537303664921466,2.242984547906586e-05,model better,model better,False,"Cross-platform transfer vs constant, intersection features",0.00017943876383252688,True,True,True,False
ARCHER2 -> Cirrus-fixed: MLP,191,1.2239078709934073,1.2126366405871438,1.2345105462387185,1.2018158544916426,1.1969564527874303,1.2159444008831113,-0.003251445517859164,-0.012909261537861871,0.013756028557226019,-0.01641013157721649,-0.09537293535526395,0.006214391884400401,98,93,-0.11638307155322862,0.16308447420800776,model better,model worse,True,"Cross-platform transfer vs constant, intersection features",0.3261689484160155,False,False,False,False
Cirrus-fixed -> ARCHER2: RF,191,1.1513926223496258,1.142688028044278,1.167881245505168,1.1183655644576789,1.097420349281956,1.1457990476584619,0.055333740779221596,0.038271710326470254,0.07194462284210679,0.04496386840332267,0.021757983417277665,0.06294059659114974,59,132,0.2955933682373473,0.00039637343011006637,model worse,model worse,False,"Cross-platform transfer vs constant, intersection features",0.0019818671505503318,True,True,True,False
Cirrus-fixed -> ARCHER2: GBQ,191,1.1602515090838297,1.1455559628741767,1.1667621516749578,1.1183655644576789,1.100109859978499,1.1438485512965912,0.055702466500704206,0.03725313541920605,0.073653202683571,0.012210621261916299,-0.02254961970510599,0.042911569479438444,67,124,0.06228184991273997,0.45542227651700373,model worse,model worse,False,"Cross-platform transfer vs constant, intersection features",0.45542227651700373,False,False,False,False
Cirrus-fixed -> ARCHER2: MLP,191,1.1624879615048411,1.1488971909632364,1.176600823300914,1.1183655644576789,1.097420349281956,1.1438485512965912,0.058859503681520486,0.025239429281407322,0.06766316415916007,0.03240991922362718,0.011403384460476104,0.05426197729640382,73,118,0.21836823734729494,0.008870745879360665,model worse,model worse,False,"Cross-platform transfer vs constant, intersection features",0.03319020400185546,True,True,True,False
ARCHER2 -> Cirrus-weak: RF,162,1.182151748537794,1.1719409609573659,1.189157463568998,1.1902011929268101,1.1741358964731683,1.2017252759034065,-0.015779916143866424,-0.023993436743838897,-0.007635380562476368,-0.013021359778981778,-0.020758250766687858,-0.003509210612471008,105,57,-0.23911232295690374,0.008297551000463865,model better,model better,False,"Cross-platform transfer vs constant, intersection features",0.03319020400185546,True,True,True,False
ARCHER2 -> Cirrus-weak: GBQ,162,1.1763614006239123,1.1670228498326596,1.1854925335974646,1.1902011929268101,1.1741358964731683,1.2017252759034065,-0.02257525156125728,-0.02794125027751493,-0.012612994990859772,-0.0158208913297605,-0.023128118762010826,-0.008611362922151977,109,53,-0.3466636370521851,0.00012969091091135032,model better,model better,False,"Cross-platform transfer vs constant, intersection features",0.0009078363763794523,True,True,True,False
ARCHER2 -> Cirrus-weak: MLP,162,1.2106720973219192,1.2003717159397285,1.2176423388904138,1.1902011929268101,1.1741358964731683,1.201571438397889,0.016872808179566823,0.005617840974898147,0.022937208475229287,0.017667148030144242,0.009660805539232542,0.025496025661585077,61,101,0.4011966977202151,9.46400182561339e-06,model worse,model worse,False,"Cross-platform transfer vs constant, intersection features",8.517601643052052e-05,True,True,True,False
Cirrus-weak -> ARCHER2: RF,191,1.1433156836124896,1.136772739758014,1.153098947043286,1.1718536799628398,1.153793972376152,1.2007697002885336,-0.0209214452079165,-0.045166197595066127,-0.006677911579041185,-0.05938973368751349,-0.09998086918496696,-0.0338047260657768,117,74,-0.5177792321116929,5.461555542777993e-10,model better,model better,False,"Cross-platform transfer vs constant, intersection features",5.461555542777993e-09,True,True,True,False
Cirrus-weak -> ARCHER2: GBQ,191,1.1472531465314029,1.1348926166935873,1.161582063136855,1.1718536799628398,1.1545008725932233,1.2007697002885336,-0.028059096734818878,-0.05844591921851183,-0.01683889234171443,-0.06302875667964408,-0.11232547203320775,-0.03746186254568124,122,69,-0.5386125654450262,1.0828504001328951e-10,model better,model better,False,"Cross-platform transfer vs constant, intersection features",1.1911354401461846e-09,True,True,True,False
Cirrus-weak -> ARCHER2: MLP,191,1.1392281017217465,1.1256888818789763,1.1523650649000652,1.1718536799628398,1.153793972376152,1.2007697002885336,-0.0778736685829009,-0.10983910365750704,-0.03812422596378795,-0.11069317667888778,-0.15883463457188096,-0.07980644135810236,122,69,-0.6115837696335078,2.3115431572289615e-13,model better,model better,False,"Cross-platform transfer vs constant, intersection features",2.7738517886747537e-12,True,True,True,False