Survival Analysis
Statistical method for analyzing time until an event occurs (e.g., churn, conversion, failure), accounting for censored data.
Survival Analysis models "When does an event happen?" (churn, conversion) and handles censored data – more powerful than binary classification for temporal questions.
Explanation
Core methods: Kaplan-Meier curves (non-parametric), Cox Proportional Hazards (semi-parametric), Accelerated Failure Time models. Censoring means: We know the event has NOT YET occurred, but not when it will.
Marketing Relevance
Ideal for marketing: Customer lifetime modeling, churn prediction with temporal context, trial-to-paid conversion timing, retention analysis.
Example
Kaplan-Meier curve shows: 50% of free trial users convert within 7 days, then the curve flattens – insight for trigger email timing.
Common Pitfalls
Cox model assumes proportional hazards (often violated). Incorrectly modeling censoring leads to bias. Ignoring competing risks distorts results.
Origin & History
Kaplan & Meier published their estimator in 1958. Cox Proportional Hazards (1972) became the most-cited statistics paper. Today standard in customer analytics, medicine, and reliability engineering.
Comparisons & Differences
Survival Analysis vs. Logistic Regression
Logistic regression predicts "whether" an event occurs; Survival Analysis predicts "when" – with censoring and temporal dynamics.
Survival Analysis vs. Time Series Analysis
Time series analyzes values over time; Survival Analysis analyzes time until a specific event.