Note On Computing Loss Function For Normal Demand

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Note on computing loss function for Normal Distribution

In this note we describe how to compute the expected unmet demand (expected shortfall) when
demand is normally distributed.
Notation:
Q: inventory quantity, a constant
D: demand, a random variable
X: sales, equal to the amount fo demand met with inventory, a random variable
S: shortfall, or unmet demand, a random variable
R: inventory leftover after meeting demand, or remaining inventory, a random variable

As presented in the video, we have the following relationships between the random variables:

X = min [Q, D ]
S= D − X
R= Q − X
From this we can find the expectations:

[ D] E [ X ] + E [ S ]
E=
=Q E [ X ] + E [ R]

If we are given the inventory constant Q and the expectation of demand, e.g., E [ D ] = µ , then by

computing E [ S ] , we also get E [ X ] , E [ R ] .

Now suppose that demand D is normally distributed with mean µ and standard deviation σ. Then

we can express the expected shortfall as:


E [=
S] ∫ ( x − Q ) φ ( x | µ , σ )dx
Q

where φ ( x | µ , σ ) is the probability density function for a normally-distributed random variable with

mean µ and standard deviation σ. As a special property of the normal distribution, one can re-write the

above expression (after some algebraic manipulations) in terms of the standard normal random
variable:

1
∞ ∞
E [S ] =
∫ ( x − Q ) φ ( x | µ , σ )dx =
σ ∫ ( x − z ) φ ( x )dx (1)
Q z

Q−µ
where z= and φ ( x ) denotes the probability density function for a standard normal random
σ
variable, i.e., mean µ = 0 and standard deviation σ = 1.

To evaluate (1) we use the following result (which requires even more arduous algebraic manipulations):

x =∞

∫ ( x − z ) φ ( x=
) dx φ ( z ) − z × (1 − Φ ( z ) )
x= z

(2)
NORMDIST ( z , 0,1, FALSE ) − z × (1 − NORMDIST ( z , 0,1, TRUE ) )

where Φ ( z ) is the cumulative distribution function for the standard normal. The spreadsheet

commands are shown above for this calculation.

The above integral in (2) is known as the loss function or sometimes, the partial loss function.

You might also like