Exponential Distribution — Time Between Events

Foundations of Statistics

Modeling Waiting Times and Survival

The exponential distribution is the mathematical backbone for modeling time between random events, from customer arrivals to equipment failures. Its unique memoryless property makes it indispensable for reliability engineering and queueing theory.

Reliability Engineering — Predicting time until component failure in aerospace and electronics
Telecommunications — Modeling call arrivals and service times in network traffic analysis
Healthcare — Estimating waiting times between patient arrivals in emergency departments

The exponential distribution tells us how long we must wait for the next event.

Core Concepts

The exponential distribution models the waiting time between consecutive events in a Poisson process. It is the continuous analog of the geometric distribution and the only continuous distribution that possesses the memoryless property.

DfExponential Distribution

An exponential random variable $X$ models the time between events occurring at a constant rate $\lambda > 0$ . Written $X \sim \text{Exp}(\lambda)$ with pdf:

f(x) = \lambda e^{-\lambda x}, \quad x \geq 0

Equivalently, $X$ is the waiting time for the first event in a Poisson process with rate $\lambda$ .

PDF of Exponential Distribution

f(x) = \lambda e^{-\lambda x}, \quad x \geq 0

Here,

$\lambda$ =Rate parameter (events per unit time)
$1/\lambda$ =Mean waiting time
$x$ =Time (non-negative)

Proof of the Memoryless Property

ThThe Exponential is the Only Memoryless Continuous Distribution

The exponential distribution satisfies the memoryless property:

P(X > s + t \mid X > s) = P(X > t), \quad \forall \, s, t \geq 0

Proof:

P(X > s + t \mid X > s) = \frac{P(X > s + t)}{P(X > s)} = \frac{e^{-\lambda(s+t)}}{e^{-\lambda s}} = e^{-\lambda t} = P(X > t)

Uniqueness: Suppose a continuous $X > 0$ satisfies $P(X > s+t \mid X > s) = P(X > t)$ for all $s, t \geq 0$ . Let $G(t) = P(X > t)$ (the survival function). The memoryless property gives:

\frac{G(s+t)}{G(s)} = G(t) \implies G(s+t) = G(s)G(t)

This is Cauchy's functional equation with solution $G(t) = e^{-\lambda t}$ for some $\lambda > 0$ (using continuity/monotonicity). Thus $F(t) = 1 - e^{-\lambda t}$ , which is exponential.

Interpretation

If you've already waited 10 minutes for a bus that arrives exponentially with rate $\lambda$ , the probability you wait another 5 minutes is the same as if you just arrived. The process "forgets" your waiting time.

Interactive Visualization

Exponential Distribution — Interactive Explorer

Mean (μ) = 1.0000Var = 1.0000σ = 1.0000

Effect of Rate Parameter on Exponential Distribution

Mean (μ) = 1.0000Var = 1.0000σ = 1.0000

CDF and Survival Function

F(x) = 1 - e^{-\lambda x}, \quad S(x) = P(X > x) = e^{-\lambda x}

Here,

$\lambda$ =Rate parameter
$S(x)$ =Survival (reliability) function

Mean, Variance, and Higher Moments

ThDerivation of Mean and Variance

Mean:

E[X] = \int_0^{\infty} x \cdot \lambda e^{-\lambda x} \, dx = \lambda \cdot \frac{1}{\lambda^2} = \frac{1}{\lambda}

(Integration by parts: $\int_0^\infty x e^{-\lambda x} dx = 1/\lambda^2$ .)

Second moment:

E[X^2] = \int_0^{\infty} x^2 \cdot \lambda e^{-\lambda x} \, dx = \lambda \cdot \frac{2}{\lambda^3} = \frac{2}{\lambda^2}

Variance:

\text{Var}(X) = E[X^2] - (E[X])^2 = \frac{2}{\lambda^2} - \frac{1}{\lambda^2} = \frac{1}{\lambda^2}

Exponential Mean and Variance

E[X] = \frac{1}{\lambda}, \quad \text{Var}(X) = \frac{1}{\lambda^2}

Here,

$\lambda$ =Rate parameter
$1/\lambda$ =Mean time between events

Coefficient of Variation

The standard deviation equals the mean: $\sigma = 1/\lambda = E[X]$ . Therefore the coefficient of variation $CV = \sigma/\mu = 1$ for all exponentials, regardless of $\lambda$ . This is a useful diagnostic: if data has $CV \approx 1$ , the exponential may be appropriate.

MGF and Moments

MGF of Exponential Distribution

M_X(t) = \frac{\lambda}{\lambda - t}, \quad t < \lambda

Here,

$t$ =Transform variable (must be less than λ)

Why t < λ

The MGF $M_X(t) = \int_0^\infty e^{tx} \lambda e^{-\lambda x} dx = \lambda \int_0^\infty e^{-(\lambda-t)x} dx$ diverges for $t \geq \lambda$ , so the MGF only exists for $t < \lambda$ .

ThDerivation of nth Moment

Differentiating $M_X(t) = \lambda(\lambda - t)^{-1}$ $n$ times:

M_X^{(n)}(t) = \frac{\lambda \cdot n!}{(\lambda - t)^{n+1}}

At $t = 0$ : $E[X^n] = M_X^{(n)}(0) = \frac{n!}{\lambda^n}$ . In particular: $E[X] = 1/\lambda$ , $E[X^2] = 2/\lambda^2$ , $E[X^3] = 6/\lambda^3$ .

Hazard Rate (Failure Rate)

Hazard Rate of the Exponential

h(x) = \frac{f(x)}{S(x)} = \frac{\lambda e^{-\lambda x}}{e^{-\lambda x}} = \lambda

Here,

$h(x)$ =Hazard rate at time x
$\lambda$ =Constant rate parameter

ThConstant Hazard Rate Characterizes the Exponential

The hazard rate $h(x) = f(x)/S(x)$ measures the instantaneous failure rate. For the exponential, $h(x) = \lambda$ is constant — the probability of failure in the next instant doesn't depend on how long the component has survived. This is the physical interpretation of the memoryless property.

Uniqueness: Any continuous distribution with constant hazard rate must be exponential.

Relationship to the Poisson Process

ThPoisson Process and Exponential Inter-Arrivals

Consider events occurring at rate $\lambda$ in a Poisson process. Then:

The number of events in time $t$ is $N(t) \sim \text{Poisson}(\lambda t)$
The inter-arrival times $X_1, X_2, \ldots$ are i.i.d. $\text{Exp}(\lambda)$
The waiting time for the $n$ -th event is $W_n = X_1 + \cdots + X_n \sim \text{Gamma}(n, \lambda)$

Proof of (2): $P(X_1 > t) = P(N(t) = 0) = e^{-\lambda t}$ (no events in $[0,t]$ ), so $F_{X_1}(t) = 1 - e^{-\lambda t}$ , which is $\text{Exp}(\lambda)$ . For subsequent inter-arrivals, use the memoryless property: after the $k$ -th event, the process "restarts."

The Poisson-Exponential Connection

Two Ways to Characterize the Poisson Process

A Poisson process with rate $\lambda$ can be defined equivalently by:

Counting definition: $N(t) \sim \text{Poisson}(\lambda t)$ , independent increments
Timing definition: Inter-arrival times are i.i.d. $\text{Exp}(\lambda)$

These two characterizations are mathematically equivalent and lead to different proofs of the same results.

Worked Example

Example: Server Request Processing

A web server receives requests at rate $\lambda = 5$ per minute. What is the probability that the time between two consecutive requests exceeds 30 seconds?

P(X > 0.5) = e^{-5 \times 0.5} = e^{-2.5} \approx 0.082

What is the probability that more than 1 minute elapses between requests?

P(X > 1) = e^{-5} \approx 0.0067

The median waiting time: solve $e^{-\lambda m} = 0.5 \implies m = \ln(2)/\lambda = \ln(2)/5 \approx 0.139$ minutes $\approx 8.3$ seconds. Note: the median is always less than the mean for exponential distributions.

Gamma Distribution Connection

ThExponential as a Special Case of the Gamma

If $X \sim \text{Exp}(\lambda)$ , then $X \sim \text{Gamma}(1, \lambda)$ (shape 1, rate $\lambda$ ).

More generally, the sum of $n$ i.i.d. $\text{Exp}(\lambda)$ variables is $\text{Gamma}(n, \lambda)$ :

X_1 + \cdots + X_n \sim \text{Gamma}(n, \lambda)

This follows from the MGF: $(M_X(t))^n = \left(\frac{\lambda}{\lambda-t}\right)^n = M_{\text{Gamma}(n,\lambda)}(t)$ .

Specific Applications

Queueing theory — Service times in M/M/1 queues are $\text{Exp}(\mu)$ ; inter-arrival times are $\text{Exp}(\lambda)$ .
Reliability engineering — Component lifetime before random failure (constant failure rate).
Physics — Radioactive decay: time until a particle decays is $\text{Exp}(\lambda)$ where $\lambda$ is the decay constant.
Network traffic — Inter-packet arrival times in Poisson traffic models.

Key Takeaways

Summary: Exponential Distribution

Models time between events in a Poisson process
PDF: $f(x) = \lambda e^{-\lambda x}$ , Mean: $1/\lambda$ , Variance: $1/\lambda^2$
Memoryless property: $P(X > s+t \mid X > s) = P(X > t)$ — the only continuous distribution with this property
Constant hazard rate $h(x) = \lambda$ — characterizes memorylessness
MGF: $M_X(t) = \lambda/(\lambda - t)$ for $t < \lambda$ ; $E[X^n] = n!/\lambda^n$
Special case of Gamma ( $\alpha=1$ ); sum of $n$ exponentials is $\text{Gamma}(n, \lambda)$
Poisson connection: inter-arrival times of Poisson process are i.i.d. exponential

Exponential Distribution — Time Between Events

Exponential Distribution — Time Between Events

Modeling Waiting Times and Survival

Core Concepts

DfExponential Distribution

PDF of Exponential Distribution

Proof of the Memoryless Property

ThThe Exponential is the Only Memoryless Continuous Distribution

Interactive Visualization

CDF and Survival Function

CDF and Survival Function

Mean, Variance, and Higher Moments

ThDerivation of Mean and Variance

Exponential Mean and Variance

MGF and Moments

MGF of Exponential Distribution

ThDerivation of nth Moment

Hazard Rate (Failure Rate)

Hazard Rate of the Exponential

ThConstant Hazard Rate Characterizes the Exponential

Relationship to the Poisson Process

ThPoisson Process and Exponential Inter-Arrivals

The Poisson-Exponential Connection

Worked Example

Gamma Distribution Connection

ThExponential as a Special Case of the Gamma

Specific Applications

Key Takeaways

Summary: Exponential Distribution

Premium Content

Need Expert Statistics Help?