--- id: P-REINFORCE-AUTO-DSUX-001 category: "10_Wiki/πŸ’‘ Topics/AI" confidence_score: 0.94 tags: [auto-reinforced, data-science, ux, user-experience, quantitative-analysis, a-b-testing, behavioral-data] last_reinforced: 2026-04-20 --- # [[Data-Science-in-UX|Data-Science-in-UX]] ## πŸ“Œ ν•œ 쀄 톡찰 (The Karpathy Summary) > "κ²½ν—˜μ˜ μ •λŸ‰ν™”: 'μ‚¬μš©μžκ°€ μ’‹μ•„ν•˜λŠ” 것 κ°™λ‹€'λŠ” 주관적 λŠλ‚Œ λŒ€μ‹ , 수백만 건의 클릭 λ‘œκ·Έμ™€ μž”λ₯˜ μ‹œκ°„ 데이터λ₯Ό λΆ„μ„ν•˜μ—¬ μ–΄λ–€ λ””μžμΈμ΄ μ§„μ§œλ‘œ μ‚¬μš©μžμ˜ κ°€μΉ˜λ₯Ό λ†’μ˜€λŠ”μ§€ 숫자둜 증λͺ…ν•˜λŠ” λ””μžμΈ μ‹¬νŒκ΄€." ## πŸ“– κ΅¬μ‘°ν™”λœ 지식 (Synthesized Content) UX 데이터 κ³Όν•™(Data-Science-in-UX)은 λŒ€κ·œλͺ¨ μ‚¬μš©μž 행동 데이터λ₯Ό μˆ˜μ§‘ν•˜κ³  λΆ„μ„ν•˜μ—¬ μ‚¬μš©μž κ²½ν—˜μ„ κ°œμ„ ν•˜κΈ° μœ„ν•œ 데이터 기반 λ””μžμΈ λ°©λ²•λ‘ μž…λ‹ˆλ‹€. 1. **3λŒ€ 뢄석 기법**: * **A/B Testing**: 두 κ°€μ§€ μ‹œμ•ˆ 쀑 μ–΄λ–€ 것이 λͺ©ν‘œ μ§€ν‘œ(클릭λ₯ , ꡬ맀 μ „ν™˜ λ“±) κ°œμ„ μ— νš¨κ³Όμ μΈμ§€ μ‹€ν—˜. * **Cohort Analysis**: νŠΉμ • μ‹œκΈ°μ— μœ μž…λœ μ‚¬μš©μž 그룹의 μœ μ§€μœ¨ 및 행동 νŒ¨ν„΄ 좔적. * **Funnel Analysis**: μ‚¬μš©μžκ°€ 각 단계(Touchpoint)μ—μ„œ μ–Όλ§ˆλ‚˜ μ΄νƒˆν•˜λŠ”μ§€ 병λͺ© ν˜„μƒ νŒŒμ•…. (Customer-Journey-Mapping와 μ—°κ²°) 2. **μ™œ μ€‘μš”ν•œκ°€?**: * λ””μžμ΄λ„ˆμ˜ 직관(Intuition)κ³Ό λ°μ΄ν„°μ˜ 객관성(Data-driven) μ‚¬μ΄μ˜ 가ꡐ 역할을 ν•˜μ—¬, κ°€μž₯ 효과적인 μ œν’ˆ κ°œμ„  μš°μ„ μˆœμœ„λ₯Ό κ²°μ •ν•˜κΈ° λ•Œλ¬Έμž„. (Priority와 μ—°κ²°) ## ⚠️ λͺ¨μˆœ 및 μ—…λ°μ΄νŠΈ (Contradictions & RL Update) - **κ³Όκ±° λ°μ΄ν„°μ™€μ˜ 좩돌**: κ³Όκ±°μ—λŠ” "μˆ«μžλŠ” 감정을 μ„€λͺ…ν•˜μ§€ λͺ»ν•œλ‹€"λ©° μ •μ„± 쑰사(Qualitative)만 μ€‘μ‹œν–ˆμœΌλ‚˜, ν˜„λŒ€ 정책은 μˆ«μžκ°€ λ§ν•΄μ£ΌλŠ” 'ν˜„μƒ μ •μ±…'κ³Ό 인터뷰가 λ§ν•΄μ£ΌλŠ” '이유'λ₯Ό κ²°ν•©ν•œ '믹슀-λ©”μ†Œλ“œ(Mixed Methods) μ •μ±…'이 ν‘œμ€€μ΄ 됨(RL Update). (Scientific-Method와 μ—°κ²°) - **μ •μ±… λ³€ν™”(RL Update)**: μ΄μ œλŠ” λ‹¨μˆœ 톡계 뢄석 정책을 λ„˜μ–΄, AI κ°€ μ‚¬μš©μžμ˜ μ‹€μ‹œκ°„ 감정 μ •μ±…μ΄λ‚˜ 뢈만쑱 정책을 μ˜ˆμΈ‘ν•˜μ—¬ μ„ μ œμ μœΌλ‘œ UIλ₯Ό λ³€κ²½ν•˜λŠ” 'μ˜ˆμΈ‘ν˜• μΈν„°νŽ˜μ΄μŠ€ μ •μ±…(Predictive UI)'으둜 μ§„ν™” μ€‘μž„. ## πŸ”— 지식 μ—°κ²° (Graph) - [[Customer-Journey-Mapping|Customer-Journey-Mapping]], [[Scientific-Method|Scientific-Method]], Priority, [[Efficiency|Efficiency]], [[Analysis|Analysis]], [[Sensitivity-Analysis|Sensitivity-Analysis]] - **Key Concepts**: HEART framework (Google), North Star Metric. ---