Kaplan–Meier Estimator
Definition
The Kaplan-Meier estimator estimates a survival function from time-to-event data and can account for right-censored observations.
It calculates the probability of not having experienced an event by a certain time as the product of conditional survival probabilities. Censoring must be correctly modeled.
Simply explained
How to estimate how many cases remain event-free after various time points, even if some observations end earlier.
Why it matters in practice
The estimator supports the study of time intervals until an event, such as failure, cancellation, or required maintenance.
Example from business
A company investigates how long machines run until the first critical failure, although not all machines fail during the observation period.
Limits and pitfalls
- Ignoring censoring; comparing groups without testing; overlooking competing risks; extrapolating curves beyond observed data.
Origin & history
Kaplan and Meier published the product-limit estimator for incomplete observations in 1958.
Use in mid-sized companies
- Possible applications: downtime, customer retention, repair duration, and other time-to-event questions.
Frequently asked questions
What does right-censored mean?
The event has not occurred by the end of observation; the exact later event time remains unknown. Is Kaplan-Meier causal? No.
Sources
- Nonparametric Estimation from Incomplete Observations · Kaplan & Meier (1958), Nonparametric Estimation from Incomplete Observations
- Survival Analysis, Abschnitt 1.2 · SAS, Survival Analysis, Abschnitt 1.2
- Kaplan-Meier-Implementierung · SAS, Kaplan-Meier-Implementierung
Source verification: 10./11.10.2026.
Editorially reviewed on: