การทดสอบแบบปรับตัว A/B — การทดสอบแบบปรับตัว A/B
การทดสอบแบบปรับตัว A/B คือการออกแบบการทดลองที่จัดสรรการเข้าชมหรือผู้เข้าร่วมการทดลองไปยังรูปแบบที่ทำงานได้ดีกว่าแบบไดนามิกในระหว่างการทดลองเอง แทนที่จะคงการจัดสรรไว้คงที่จนกว่าจะสิ้นสุด โดยอาศัยอัลกอริทึม multi-armed bandit เช่น Thompson Sampling หรือ Upper Confidence Bound (UCB) การทดสอบนี้จะสร้างสมดุลระหว่างการสำรวจรูปแบบที่ไม่แน่นอนกับการใช้ประโยชน์จากรูปแบบที่แสดงประสิทธิภาพที่เหนือกว่าอยู่แล้ว ซึ่งโดยทั่วไปจะให้ผลลัพธ์รวมที่สูงขึ้น ในขณะเดียวกันก็ยังคงให้ข้อสรุปเชิงอนุมานที่ถูกต้อง
อ่านวิธีฉบับเต็ม
เข้าสู่ระบบด้วยบัญชีฟรีเพื่ออ่านส่วนนี้
Method map
The neighbourhood of related methods — select a node to explore.
แหล่งอ้างอิง
- 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 ↗
- 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 ↗
วิธีอ้างอิงหน้านี้
ScholarGate. (2026, June 3). Adaptive A/B Testing. ScholarGate. https://scholargate.app/th/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.
- การออกแบบแบบ ABการออกแบบการทดลอง↔ compare
- การทดลองแบบปรับตัวการออกแบบการทดลอง↔ compare
- Blocked A/B Testการออกแบบการทดลอง↔ compare
- การทดสอบ A/B แบบแฟกทอเรียลการออกแบบการทดลอง↔ compare
- การทดลองแบบหลายแขนการออกแบบการทดลอง↔ compare
- การทดลองสุ่มและมีกลุ่มควบคุม (Randomized Controlled Trial - RCT)การออกแบบการทดลอง↔ compare