- exponential distribution: X = time until the next event, X≥0, with λ the average rate of events
- continuous cousin of waiting for the first success, and the only continuous memoryless distribution
fX(x)=λe−λx,FX(x)=1−e−λx,x≥0
- $E(X) = \frac{1}{\lambda}$, $Var(X) = \frac{1}{\lambda^2}$, $SD(X) = \frac{1}{\lambda}$
- tail: $P(X > x) = e^{-\lambda x}$
- example: starting at 9:00pm, the wait for the next text averages 5 minutes
- E(X)=5=λ1, so λ=51 and FX(x)=∫0x51e−t/5dt=1−e−x/5
- no text in the next 10 minutes: P(X>10)=1−F(10)=e−2=0.1353
- text between 9:07 and 9:10: P(7<X<10)=F(10)−F(7)=e−1.4−e−2=0.111
- no text by 9:05, chance it arrives by 9:10: Conditional Probability with A={X<10}, B={X>5}
P(X<10∣X>5)=P(X>5)P(5<X<10)=1−F(5)F(10)−F(5)=e−1e−1−e−2=1−e−1=0.632
- which is just $P(X < 5)$: the 5 minutes already waited don't matter, that's [[Memoryless Property For Exponentials|memorylessness]]