The Poisson Distribution

1
 Poisson distribution
 The outcome of a random variable with a Poisson distribution is discrete
(0,1,2,3,…) and positive (i.e., no negative outcomes).
 The Poisson is characterized by a single parameter, its mean:
 Recall, the Normal distribution is defined by two parameters, the mean
and standard deviation.
 The standard deviation of a Poisson distribution is equal to the square
root of its mean.
 The Poisson is related to the exponential distribution:
 If the inter-arrival times of customers are exponentially distributed, then
the number of customer arrivals in a fixed time interval is Poisson
distributed.
 The Poisson is ideal for describing the demand of slow moving products,
e.g., products that have average sales of 10 or fewer units over a particular
period of time.
2
 The shape of the Poisson distribution
 The Poisson’s density function takes on different shapes, depending on its
mean
 With a large mean (more than 10) the Poisson takes on a “bell shape”
 With a small mean (less than 5) the Poisson takes on different shapes
that slowly change in the direction of a bell curve
0.60
0.20
0.18 5
0.50
Density function probability

0.625 0.16

Density function probability

0.40 0.14
10
1.25 0.12
0.30 0.10 20
0.08
0.20 2.5
0.06

0.10 0.04
0.02
0.00 0.00
0 2 4 6 8 10 0 10 20 30
Q Q

Density function of 6 Poisson distributions with means

3 0.625, 1.25, 2.5, 5, 10 and 20
 Poisson’s Distribution and Loss Functions
 There is no equivalent of the “z-statistic” for the Poisson like there is for the
Normal distribution.
 Hence, you need a distribution and loss function table for each Poisson
distribution.
Q f (Q ) F (Q ) L(Q)
 Consider the density function, f(Q), 0 0.286505 0.286505 1.25000
distribution function, F(Q), and the 1 0.358131 0.644636 0.53650
loss function, L(Q) of the Poisson with 2 0.223832 0.868468 0.18114
mean 1.25 in the table to the right: 3 0.093263 0.961731 0.04961
 What is the probability that 4 0.029145 0.990876 0.01134
demand is exactly 2? 22.38% 5 0.007286 0.998162 0.00221
 What is the probability that 6 0.001518 0.999680 0.00038
demand is no more than 3? 7 0.000271 0.999951 0.00006
96.17% 8 0.000042 0.999993 0.00001
 If you have 2 units to sell, then 9 0.000006 0.999999 0.00000
what is expected lost sales? 10 0.000001 1.000000 0.00000
0.1811
4
 The Order-up-to Model:
Choosing an order-up-to level, S, to
meet an in-stock service target

5
 Calculating demand over L+1 periods:
Poisson distributed demand

 Assume that the period length is 1 day and the lead time is L
days.

 Assume that daily demand is Poisson with mean equal to λ

 Then,
 L+1 demand is Poisson with…
o Mean

6
 Demand over L+1 periods in Susan’s Territory

 The period length is 1 day, and Susan’s replenishment lead

time is 1 day
16
 Daily demand is Poisson with…
14
 Mean = 0.29 12

10
 L+1 demand is Poisson with…

Units
8

 Mean = (1+1) x 0.29 = 0.58 6

Jul
Aug
Jun
May
Jan

Apr

Dec
Mar

Nov
Oct
Feb

Sep
Month

7
 Performance measures in Susan’s territory
Mean demand = 0.29 Mean demand = 0.58
S F(S) L (S ) S F(S) L (S )
0 0.74826 0.29000 0 0.55990 0.58000
1 0.96526 0.03826 1 0.88464 0.13990
2 0.99672 0.00352 2 0.97881 0.02454
3 0.99977 0.00025 3 0.99702 0.00335
4 0.99999 0.00001 4 0.99966 0.00037
5 1.00000 0.00000 5 0.99997 0.00004
F (S ) = Prob {Demand is less than or equal to S}
L (S ) = loss function = expected backorder = expected
amount demand exceeds S
 Suppose Susan operates with S = 3:
– What is the expected backorder? 0.00335
– What is the in-stock probability? 99.7%
– What is the expected on-order inventory? 0.29
– What is the expected cycle inventory? 0.29/2
– What is the expected on-hand inventory? 3 – 0.58 + 0.00335
8 – What is the expected fill rate? 1 – 0.00335/0.29
 Choosing S to meet a target in-stock with Poisson
demand
Mean demand = 0.29 Mean demand = 0.58
S F(S) L (S ) S F(S) L (S )
0 0.74826 0.29000 0 0.55990 0.58000
1 0.96526 0.03826 1 0.88464 0.13990
2 0.99672 0.00352 2 0.97881 0.02454
3 0.99977 0.00025 3 0.99702 0.00335
4 0.99999 0.00001 4 0.99966 0.00037
5 1.00000 0.00000 5 0.99997 0.00004
F (S ) = Prob {Demand is less than or equal to S}
L (S ) = loss function = expected backorder = expected
amount demand exceeds S

– What S should Susan choose to yield an in-stock probability of

at least 99%? 3
– What if the target is 99.9%? 4
– What if the target is 99.99%? 5
9
 The Order-up-to Model:
General case

10
 Periodic Review Policy: OUT-level and safety Inventory
for Normally distributed demand
 Consider the general case where:
− T : Reorder interval
− L : Lead time
− μ, σ : mean and standard deviation of demand per unit time
(assumed normally distributed)

 Then, the order-up-to level and safety cost calculations can be

generalized as follows:

11