Courbes ROC et AUC
Courbe ROC dafay wane ni classifier bi di tàqalee ñaari class ci bepp dogal bu mëna am, AUC dafay comprimé courbe bi yépp ci benn nimero.
Résumé
Together they tell you ranking quality independent of where you draw the cutoff.
Plongeur bu xóot
Benn courbe buy màndargaal ni Reseptër bi di doxee (ROC) dafay wane Taux Positif Dëgg (sensitivite, ci axe y) ak Taux Positif Njuumte (1 dindi ci specificité, ci axe x) sooy gliise threshold classification bi daale ko ci 1 wàcci ba 0. Threshold bu nekk dafay joxe benn poñ; boole leen dafay topp courbe bi. Benn model buy rang bepp positif ci kaw bepp negatif dafay laxasu koñu kaw-cammoy bi. Barab bi ci suufu courbe bi (AUC) mooy natt barab bi nekk ci suufu liiñ bii, dalee ko ci 0.5 (xalaat ci anam wu bari, diagonal bi) ba ci 1.0 (baax). Tekki bu am solo: AUC mingi méngoo ak probabilite bi model bi am poñ yu bari yuñ tànnee ci anam wu mucc, moo gëna bari ci negatif yiñ tànnee ci anam wu mucc. Tur wi mingi cosaanoo ci operatëri radar yu ñaareelu guerre mondiale di wuutale siñaal ak bruit.
Gis-gis xarala
AUC moom boppam ci threshold ndax dafay boole performance ci bépp dagg, kon amul benn njeexte ci fi nga def dogal bi. Ci wàllu math mingi méngoo ak lim bu Mann-Whitney U ak test rank-sum bu Wilcoxon, loolu dafay tekki ni mingi aju ci rang yi ñu séentu ci poñ yi, te baña aju ci seeni valeur absolu. Loolu daf koy def mu dëgër ci biir coppite poñ yu monotonik waaye itam du sensitif ci etalonnage: xeetu rang bu baax waaye bu baaxul ci etalonnage mën na ba leegi am AUC bu yéeg.
njeextalu pexe
dogal yu gëna leer
Daf lay jàppale nga tàqale kàddu yu leer ci wàllu xarala ak làkku fësal njaay.
Njëgg ak budget
Mën nga laaj laaj yu gëna baax ci samp gi balaa ngay dugal xaalis wala sa jotu liggéey.
Ekip ak def liggéey
Ekip yi bokk xam-xam ñoo gëna mëna jël yenn dogal ci wàllu produit, politik ak jàng.
Ëlëgu Courbes ROC ak AUC
ROC-AUC desna nekk metric rapoor buñ jagleel, waaye praktiseur yi dañu koy gëna boole ak curves Precision-Recall ngir done yu desekilibre lool, fu ROC mëna nuru lu gëmm. Xaarandi jëfandikoo bu gëna yaatu ci AUC (focus ci gox bu baax-njuumte-positif bi am solo ci liggéey bi), jàngat bu sensible ak decision-curve, ak AUC bu nekk ci subgroupe bi di rapoor ci surface justice gaps. Ginaaw model yi dañuy dundal dogal yu dëggu yi, metrics etalonnage ak AUC dina ñu gëna xamle ñoom ñaar, du AUC kese.
Doxal ci àdduna dëgg
Tegtale ñaari xeeti gis-gis njuuj njaaj ci bànk ci seen AUC ngir tànn bi gëna mëna rang jëflante yu njuuj njaaj yi ci kaw jëflante yu baax yi
Dañu jàngat test diagnostic ngir feebar (lu melni, classifier depistaasu cancer) fu radiologist yi wara jëflante ngir jàpp ñu bari ci alarm yu baaxul yi
Defar buntu filtre spam bi jëfandikoo courbe ROC ngir tëye njuumte positif yi (mail bu baax buñu màndargaale ni spam)
Benchmarking ab xeetu poñ leble-default fu AUC di tënk ni ñuy tàqalee leblekat yiy fay ak ñiy default
Risk yi ak balustrade yi
Ekip yu bari mën nañu jëfandikoo benn baat ci anam wu wuute, kon teela leeral yaatuwaayam.
Benchmark yi mën nañu nuru lu am doole waaye performance yi ci àdduna bi duñu tolloo.
Bëgg kalite done ak palaŋu jàngat dafay faral di jur njariñ yu yomba dagg.
Roadmap ngir samp gi
Tàmbaleel ci joxe leeral ci làkk wu leer ci njariñ li nga soxla.
Tannal benn metric bu baax ak benn anam bu baaxul balaa ngay saytu.
Doxal ab pilote bu ndaw ak ay done yu representatif, du ab demo bu leer.
Bindal fi ROC Curves ak AUC di jàppale ak fi pexe yu gëna yomba gëna baax.
Weyal di banneexu
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the ROC Curves and AUC quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Gis bi ci topp
Kompromis Bias-Variance
Laaj yi ñuy faral di laaj
What is ROC Curves and AUC?
Courbe ROC dafay wane ni classifier bi di tàqalee ñaari class ci bepp dogal bu mëna am, AUC dafay comprimé courbe bi yépp ci benn nimero. Ñoom ñépp ñu ngi lay wax kalite rang bi amul benn njeexte ci fi nga jëlee cutoff bi.
Ban ñaari limu courbe ROC di tracé seen biir?
Benn courbe ROC dafay wane tolluwaayu positif bu dëggu bi (sensitivite) ci axe y bi ak tolluwaayu positif bu baaxul bi (1 - mën mën) ci kaw axe x bi ci biir threshold yi.
Ban valeur AUC moo méngoo ak classificateur bi gënawul def ludul guessing random?
Benn xeetu xaaj bu bari dafay topp liiñ diagonaal bi, di joxe AUC bu 0.5. AUC bu 1.0 baaxna, te 0.0 dafay tekki ni dafay am rang bu nekk ci ginaaw.
Lu tax ñuy wax AUC ni 'moom boppam ci threshold'?
AUC dafay tënk courbe ROC bi yépp, bi ñuy génne ci bale bépp threshold, kon nekkul ci benn dagg buñ tànn.
Ngir dataset bu desekilibre bu baax, ban courbe lañuy faral di digal ñu boole ci wala ci palaasu ROC ndax ROC mën na nuru lu ëpp yaakaar?
Ci jafe-jafe yu bari yu baaxul yi, limu negatif yu bari yi dafay tëye Njuumteg Positif yi, di yokk ROC. Courbe Precision-Recall dafay xoole ci klaas bu baax bi ba noppi di wane performance bu gëna dëggu.