Process / pipelineDeneysel desen

Adaptivno A/B testiranje — Adaptivno A/B testiranje

Adaptivno A/B testiranje je eksperimentalni dizajn koji dinamički preraspoređuje promet ili sudionike prema bolje performirajućim varijantama tijekom samog eksperimenta, umjesto da alokacije ostanu fiksirane do kraja. Koristeći algoritme za višeruke razbojnike (multi-armed bandit) poput Thompsonovog uzorkovanja (Thompson Sampling) ili Gornje granice pouzdanosti (Upper Confidence Bound - UCB), ono balansira istraživanje nesigurnih varijanti s iskorištavanjem onih koje već pokazuju superiorne performanse, tipično ostvarujući veće ukupne ishode uz istovremeno postizanje valjanih inferencijalnih zaključaka.

Pronađite temu uz PaperMindUskoroVideoUskoroDownload slides

Pročitajte cijelu metodu

Samo za članove

Prijavite se besplatnim računom kako biste pročitali ovaj odjeljak.

Prijavite se

Method map

The neighbourhood of related methods — select a node to explore.

Izvori

  1. Russo, D., Van Roy, B., Kazerouni, A., Osband, I., & Wen, Z. (2018). A Tutorial on Thompson Sampling. Foundations and Trends in Machine Learning, 11(1), 1–96. DOI: 10.1561/2200000070
  2. Offer-Westort, M., Coppock, A., & Green, D. P. (2021). Adaptive Experimental Design: Prospects and Applications in Political Science. American Journal of Political Science, 65(4), 826–844. DOI: 10.1111/ajps.12597

Kako citirati ovu stranicu

ScholarGate. (2026, June 3). Adaptive A/B Testing. ScholarGate. https://scholargate.app/hr/experimental-design/adaptive-ab-test

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

Compare side by side

Citirana u

ScholarGateAdaptive A/B test (Adaptive A/B Testing). Preuzeto 2026-06-15 s https://scholargate.app/hr/experimental-design/adaptive-ab-test · Skup podataka: https://doi.org/10.5281/zenodo.20539026