Survival analysis — churn is a time-to-event, not a yes/no
Survival models estimate when an event happens, correctly handling subjects who have not churned yet (censoring) — the thing a plain classifier throws away and biases itself by discarding.
Not yet scoped.
Why
The PoC fits a Kaplan-Meier curve and a Cox model to a subscription or wallet-activity dataset and shows why censoring makes the naive churn rate wrong.
How it works
Not yet built.
Where it lands in Jayverse
Wallet: model inactivity as time-to-event, not a 30-day flag. Fit a Kaplan-Meier curve to embedded-wallet activity so wallets that simply haven't churned yet aren't counted as retained or lost by a fixed cutoff.
Verex: censor active traders instead of assuming they've left. When measuring trader or market-maker retention, treat still-active accounts as censored observations rather than folding them into a naive churn rate.
Key expressions
Words and phrases from this page worth keeping, with the Korean meaning and the sentence they come from.
Expression
뜻 · 쓰이는 자리
time-to-event
사건 발생까지 걸리는 시간(으로 보는 관점) · 이탈을 예/아니오가 아니라 시점 문제로 보는 방식 · "churn is a time-to-event, not a yes/no"
censoring
중도절단 · 아직 이탈하지 않은 대상을 올바르게 처리하는 통계적 개념 · "correctly handling subjects who have not churned yet (censoring)"
throw away
내다버리다, 무시하다 · 일반 분류기가 아직 이탈 안 한 대상 정보를 버림 · "the thing a plain classifier throws away"
bias itself
스스로를 편향시키다 · 정보를 버림으로써 결과가 왜곡됨 · "biases itself by discarding"
naive
순진한, 단순한 · 중도절단을 무시한 이탈률 계산을 가리킴 · "why the naive churn rate wrong"
Kaplan-Meier curve
카플란-마이어 곡선(Kaplan-Meier curve) · 중도절단을 반영해 생존확률을 추정하는 비모수 통계 기법. "fits a Kaplan-Meier curve and a Cox model"
Cox model
콕스 비례위험모형(Cox proportional hazards model) · 생존분석에서 여러 변수가 사건 발생 위험에 미치는 영향을 추정하는 회귀모형. "fits a Kaplan-Meier curve and a Cox model"