Exponential Distribution
Download
Report
Transcript Exponential Distribution
Exponential Distribution
-‘waiting time density’
- the time until the next event for a Poisson
distribution.
-The mean number of events per unit time
is represented by λ.
-X ~ Exp(λ)
• In Poisson – the variable is the number of events
in an interval (discrete)
• In the Exp dist – the variable is the waiting time
until the next event (time is continuous).
• The pdf:
f ( x ) e x
• x ≥ 0 as time cannot be negative.
• Exp dist is consider ‘memoryless’ – the mean
waiting time can start at any moment. If you
have waited 30 mins without the next event
occuring, the mean waiting time is still 10 mins.
-The exp dist is always a decreasing function
- The mode of the exp dist is always 0
- λ is a parameter affecting the decay rate.
• The mean =
x e dx ( x)(e
0
x
0
x
)dx
1
If the mean number of events per hour is 5 then
λ = 5 and the mean waiting time will be 1/5 so 12
minutes.
• Variance =
1
2
2
( x ) f ( x)dx 2
0
- The std dev = σ =
var( x)
1
• The probability that the waiting time is a minutes or
less when X ~ Exp(λ) is
• P(X ≤ a) =
a
0
ex dx 1 ea
- Thus the probability of waiting at least a minutes is
a
a
e
P(X ≥ a) = 1 – P(X ≤ a) = 1 – ( 1 e ) =
• Median waiting time
• So
tm
tm
1
a
e dx
2 0
ln 2
An online statistics forum gets 3 postings
per randomly distributed per hour.
• A) If a posting was just made, find the
mean waiting time to the next posting.
• Soln: the mean psoting time is 1/λ which is
1/3. this has to be converted to minutes. It
is 1/3rd of an hour which is 20 mins.
• B) If a posting was made 10 minutes ago,
find the mean waiting time to the next
posting.
• Soln: ‘memoryless’, so 20 mins.
• C) Find the standrad deviation of the waiting
time to the next posting.
• Soln: μ = σ so also 20 mins.
• D) Find the probability that the waiting time will
be 30 minutes or less.
1
x
30 1
e 20 dx 0.7769
• Soln: P(X≤ 30) = 0
20
• One can also use P(X≤a) = 1 – e
30
which is 1 e 20 = 0.777
a
• E) Find the median waiting time to the next
posting
• Soln: this can be solved using:
1
1
e
2 0 20
tm
1
1
x
20
dx [e
1
x
20 tm
0
] 1 e
1
tm
20
tm
1
1 e 20 tm 20 log e 2 13.86
2
• We could also use the formula
tm
• Here
ln 2
ln 2
tm
20 ln 2 13.9
1
20