Exponential distribution model, pdf and CDF

Exponential distribution model, pdf and CDF

Spec SF1–SF2 · AQA A-Level Further Maths: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • The exponential distribution models the time between random, independent events at a constant rate λ\lambda.
  • pdf: f(x)=λe−λxf(x)=\lambda e^{-\lambda x} for x≥0x\ge0.
  • CDF: F(x)=1−e−λxF(x)=1-e^{-\lambda x}, so P(X>x)=e−λx\mathrm{P}(X>x)=e^{-\lambda x}.
  • Always use the same time unit for xx and λ\lambda.
  • To find λ\lambda or a quantile, take logarithms.

When the exponential model is used

The time between events that occur randomly, independently and at a constant average rate.

The exponential distribution models the continuous time (or distance) between events that occur:

  • at random and independently of each other, and
  • at a constant rate λ\lambda per unit time.

Examples: the time between calls to a helpline, or between breakdowns of a machine.

If events cluster, or the rate changes (for example after servicing), the model is not appropriate.

0.511.522.530.511.522.53xyλ = 1λ = 3
Exponential pdf f(x) = λe^(−λx) for λ = 1 and λ = 3: highest at x = 0 and falling steadily, with a larger λ falling faster.

Suitable

  • Time between random helpline calls
  • Time between machine breakdowns at a steady rate

Not suitable

  • Lifetime of a part that wears out with age
  • Number of calls in an hour (that is Poisson)

Which situation is best modelled by an exponential distribution?

The pdf

It starts at height λ\lambda and decays: short gaps are more likely than long ones.

If XX has parameter λ>0\lambda>0, the probability density function is

f(x)=λe−λx,x≥0,f(x)=\lambda e^{-\lambda x},\quad x\ge0,

and f(x)=0f(x)=0 for x<0x<0.

The graph starts at height λ\lambda and decreases: short gaps are more likely than long ones.

The units of λ\lambda are "per unit of time", so XX must be in the matching unit.

0.511.522.533.544.550.511.52xy(0, 1)(0, 2)λ = 1λ = 2
Exponential pdfs: a larger λ starts higher and decays faster

What is the height of the pdf at x=0x=0?

The cumulative distribution function

Integrating the pdf gives F(x)=1−e−λxF(x)=1-e^{-\lambda x}, so P(X>x)=e−λx\mathrm{P}(X>x)=e^{-\lambda x} with no integration.

Integrating the pdf from 00:

F(x)=∫0xλe−λu du=[−e−λu]0x=1−e−λx,x≥0.F(x)=\int_0^x\lambda e^{-\lambda u}\,du=\left[-e^{-\lambda u}\right]_0^x=1-e^{-\lambda x},\quad x\ge0.

Hence P(X>x)=e−λx\mathrm{P}(X>x)=e^{-\lambda x}. Differentiating FF returns f(x)=λe−λxf(x)=\lambda e^{-\lambda x}.

The cumulative distribution function gives P(X≤x)\mathrm{P}(X\le x).

0.511.522.533.544.550.20.40.60.811.2xyy = 1median (0.693, 0.5)F(x) = 1 − e^(−x)
The CDF for λ = 1 rises from 0 towards 1

What is P(X>x)\mathrm{P}(X>x) for an exponential distribution?

Calculating probabilities

Use F(x)F(x): tails are e−λae^{-\lambda a} and intervals are a difference of exponentials.

Use F(x)F(x) or integrate f(x)f(x).

  • P(X<a)=1−e−λa\mathrm{P}(X<a)=1-e^{-\lambda a}.
  • P(X>a)=e−λa\mathrm{P}(X>a)=e^{-\lambda a}.
  • P(a<X<b)=F(b)−F(a)=e−λa−e−λb\mathrm{P}(a<X<b)=F(b)-F(a)=e^{-\lambda a}-e^{-\lambda b}.

Convert the time into the same unit as λ\lambda first.

0.20.40.60.811.21.41.61.820.511.522.53xyf(x) = 3e^(−3x)
λ = 3 per hour. The shaded area from 0.2 h to 0.5 h is P(0.2 < X < 0.5) = e^(−0.6) − e^(−1.5) = 0.326.
  • P(X>a)\mathrm{P}(X>a)e−λae^{-\lambda a}
  • P(a<X<b)\mathrm{P}(a<X<b)e−λa−e−λbe^{-\lambda a}-e^{-\lambda b}

Worked example

Calls arrive at a rate of λ=3\lambda=3 per hour. Find the probability that the gap between calls is between 12 and 30 minutes.

Buses arrive at rate λ=2\lambda=2 per hour. What is the probability the gap is longer than 30 minutes?

Finding parameters and quantiles

Set up an equation of the form e−λt=ke^{-\lambda t}=k and take logarithms.

If a probability is given, solve for λ\lambda (or tt) using logarithms: e−λt=ke^{-\lambda t}=k gives t=−ln⁡kλt=\dfrac{-\ln k}{\lambda}. Keep the value positive.

For independent gaps multiply probabilities: three gaps each longer than aa have probability (e−λa)3=e−3λa\left(e^{-\lambda a}\right)^3=e^{-3\lambda a}.

1234560.20.40.60.81xy(2, 0.3)(3.83, 0.1)P(X > t)
P(X > t) = e^(−λt) with λ = 0.602. P(X > 2) = 0.3 fixes λ, and P(X > t) = 0.1 gives t = 3.83.
  • e−λt=ke^{-\lambda t}=kt=−ln⁡kλt=\dfrac{-\ln k}{\lambda}
  • Independent gapsmultiply the probabilities

Worked example

P(X>2)=0.3\mathrm{P}(X>2)=0.3. Find λ\lambda, and then find tt with P(X<t)=0.9\mathrm{P}(X<t)=0.9.

If P(X>5)=e−1\mathrm{P}(X>5)=e^{-1}, what is λ\lambda?

Try an exam question

Calls to a helpline arrive at random at a constant average rate of 33 per hour. Find the probability that the time between two successive calls is between 1212 and 3030 minutes.

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Find the probability that the time between two successive calls is less than half an hour.

1 mark, marked against the mark scheme

Revision notes