Memoryless Property of Exponential
What you'll learn: You'll discover why the exponential distribution "forgets" how long you've already been waiting—a unique property with surprising real-world consequences.
The Core Idea
The memoryless property states that for an exponential random variable, the probability of waiting an additional t time units is the same, regardless of how long you've already waited. Mathematically:
P(X > s + t | X > s) = P(X > t)
In plain English: if you've already waited s minutes for a bus, the probability you'll wait another t minutes is identical to the probability you'd have waited t minutes from the start. The distribution has "no memory" of the time already elapsed.
Why This Matters in Practice
The exponential distribution is the only continuous distribution with this property. It models truly random, memoryless processes like:
- Customer service calls arriving: If no one has called in the last 5 minutes, that doesn't make a call in the next minute more or less likely
- Radioactive decay: An atom that hasn't decayed for an hour is no more "due" to decay than a fresh atom
- Machine failures: For certain systems, past survival time doesn't reduce future failure risk
When NOT to Use Exponential
If waiting longer changes probabilities—like "the bus is more likely to arrive if it's already been 20 minutes"—then the process has memory and exponential isn't appropriate. That's when you'd consider distributions like Weibull or gamma instead.
Key Takeaway: The exponential distribution's memoryless property means past waiting time provides zero information about future waiting time—perfect for modeling truly random arrivals and events, but inappropriate when history matters.