OMT 8604 Logistics in Supply Chain Management: Master of Business Administration

OMT 8604 Logistics in Supply
Chain Management
Master of Business Administration
Understanding the nature of
SKU
• SKUs can be diverse
• Items produced and held in inventories can
differ in cost, weight, volume, color, physical
shape
• Some large manufacturing companies and
military organizations stock more than 500,000
distinct items in inventories
Diversity of
SKU’s
Ammunition
3
Milk Meat Electronic Jewels Lead
Items
Dependent Vs. Independent Demand
Independent Demand
(finished goods and spare parts)
A Dependent Demand
(components)
B(4) C(2)
D(2) E(1) D(3) F(2)

Dependent Vs. Independent
Demand
Independent Dependent Dependent
Demand Demand Demand
A (5)
B (20)
D (40)
E (?)
C (?)
D (30)
E (?)
A-B-C Classification
System
• Useful framework to control/manage
inventory systems
– Classifies items in terms of their annual usage
– Based on Pareto’s
“In any series of elements to be controlled, a selected
small fraction, in terms of number of elements, always
accounts for a large fraction, in terms of effect.”
(Pareto,1906)”
A-B-C Classification
System
• Useful framework to control/manage
inventory systems
– Class A about 5-10% of the items account for
about
50% or more of the dollar usage.
– Class B about 50% or more of the items
account for most of the remaining
50% of the dollar usage
– Class C about 40% of the items account for a
small
% of the dollar usage
ABC
Classification
This figure suggest that SKU’s in a firm’s inventory

should not be controlled to the same extent
A-B-C Classification Procedure
• Given the demand (Di) and value (vi) for

each sku i:
1. Calculate the usage, Divi, for each sku
2. Sort sku’s in decreasing order of usage (larger
values first)
3. Calculate the % usage, % of total number
of sku’s, and the respective cumulative values
4. Assign sku’s to class A, B or C
Exercis
e
Annual
Annual
SKU Demand Unit Cost ($)
Consumption
(Units)
($)
1 1100 2 2200
2 600 40 24000
3 100 4 400
4 1300 1 1300
5 100 60
6 10 25
7 100 2
8 1500 2
9 200 2 400
10 500 1
Total
Exercise
Annual Percenta Cumulati Unit Annua Percenta Cumulative
SKU Demand l Classify
g e% v e % Cost g e% %
(Units) ($) Consu
mption
($)
2 600 10.89% 10.89% 40 24000 62.75% 62.75%
5 100 1.81% 12.70% 60 6000 15.69% 78.43%
8 1500 27.22% 2 3000 7.84% 86.27%
1 1100 19.96% 2 2200 92.03%
4 1300 23.59% 83.48% 1 1300 3.40%
10 500 92.56% 1 500 1.31% 96.73%
3 100 1.81% 94.37% 4 400 1.05% 97.78%
7 100 96.19% 2 200 98.30%
9 200 3.63% 99.82% 2 400 1.05% 99.35%
6 10 100.00% 25 250 0.65% 100.00%
Total 5510 100.00% 38250 100.00%
ABC
ANALYSIS
120%
100%
80%
Cumulative Annual
60%
40%
Usage
20%
0%
0 20 40 60% 80 100 120
% % % Cumulative % % %
Units
Managing with A-B-C Classification System
• Class A items should receive the most

expert attention
• Class B items have significant usage and
number of sku’s; they can be managed with
computer software; with personalized
intervention using exception rules
• Class C items must be managed with the simplest
decision systems, with minimal managerial
attention.
Managing with A-B-C
Classification System
Hierarchical Representation of Demand
Classification System
Inventory Management Process
Forecastin
g
• “Doubt is not a pleasant condition, but
certainty is absurd.” Voltaire (1694 - 1778)
• Forecasting estimates future demand for
inventory planning purposes
• Demand uncertainty is a major cause of
inventories in a firm when compared to other
causes (i.e., supplier variability,
manufacturing variability).
Forecastin
g
• Forecast error is one of the major causes
of customer service failures;
• In turn, customer service failures is the biggest
cause of lost business.
• We will focus on short-term forecasting of
individual sku’s (appropriate for operational
management).
Forecasting
Time Series
10 Minutes Break
Individual Item Short Term Forecasting
We will study the following models:
𝑥𝑡= 𝑎 + 𝜀𝑡 𝐿𝑒𝑣𝑒𝑙 𝑀𝑜𝑑𝑒𝑙
𝑥𝑡= 𝑎 + 𝑏𝑡 + 𝜀𝑡(𝑇𝑟𝑒𝑛𝑑 𝑀𝑜𝑑𝑒𝑙)
𝑥𝑡=(𝑎 +𝑏𝑡 𝐹𝑡+ 𝜀𝑡(𝐿𝑒𝑣𝑒𝑙 𝑀𝑜𝑑𝑒𝑙)
Where 𝑥𝑡is the demand in period t; a is the level; b
is a linear trend; and 𝜀𝑡is the independent
random variation with mean 0 and variance 𝜎2
Our focus: short term forecasting of high-
volume sku’s e.g. what will be next month’s
demand of item xyz
The parameters a, b, 𝐹and 𝜎 are unknown
𝑡 22
Individual Item Short Term Forecasting
Statistical Forecasting Procedure
Given a time series of past demand

1. Select the appropriate model
1. Plot and observe the data before using it in any
mathematical algorithm
2. Exclude data that you know is not representative of the
period you are forecasting – you may add known variations
later.
2. Estimate model parameters
3. Use the model with estimated parameters to
forecast the future demand
Simple Moving Average
Consider estimating the level
parameter a in a level model
𝑥𝑡= 𝑎 + 𝜀𝑡
Notaion:
(a-hat) is the estimate of a at the end of period t
+:𝜏𝑡 is the estimate of the forecast made at the end of
period t for the period t+𝜏
Given the demand for the most recent N time periods,
the N period moving average, as of the end of period t
is:
Simple Moving Average
At the end of period t, the estimates of the level and
forecast are:
Exercis
e
Take 5 minutes to answer the following: Given the
demand for SKU TQFP 3425-98, determine the 4-month
moving average for the months February and March
2016.
Exercis
e 4 Month
Month Period Actual Demand
Moving Average
Sep 1 104
Oct 2 109
Nov 3 114
Dec 4 83
Jan 5 125
Feb 6 94
Mar 7 90
Apr 8 102
May 9 110
Jun 10 97
Jul 11 115
Aug 12 86
130
125
120
115
110
105
100
95
90
85
80
1 2 3 4 5 6 7 8 9 10 11
12
Actual Demand
4 Month Moving Average
2 Month Moving
Average
 The smaller the number of period considered, the large the effect
of the most recent periods
 If a remains unchanged, then the larger number of periods
considered, the more accurate is the estimate
 Small number of period may cause “nervousness”; while large
number of periods may cause “insensitivity” to possible changes in
the level a.
 Typical values if N range from 3 to 12 periods.
Remarks: Moving Average
• Similar procedures to estimate trends and

other demand patterns
– Pros:
• Simple
– Cons:
• Need to keep the last N-periods of data
for b each s.k.u.
• Gives equal weight to all N observations
when computing the average.
Simple Exponential Smoothing
• Widely used
• The idea is to give newer data more weight in the
forecast estimates
• We will study its application for:
– Level demand forecasting
– Trend demand forecasting
Simple Exponential
Smoothing
Where α is known as smoothing constant

The larger the value of α the more wight is given to the
recent data
Johnson and Montgomery (1974) recommend that the
value of α is chosen between 0.1 and 0.3
Only need to store the estimate of the last period for each
s.k.u
Simple Exponential Smoothing
Exercise: Take 10 minutes to compute the March forecast
using exponential smoothing
Actual
Month Period Forecast
Demad
0 - 102.7
sep 1 104 103.22
oct 2 109
nov 3 114
dec 4 83
jan 5 125
feb 6 94
mar 7 90
apr 8 102
may 9 110
jun 10 97
jul 11 115
aug 12 86
1. Use average demand for first four period as a0 =102.7
2. Use α =0.4
Simple Exponential
Smoothing
130
125
120
115
110
105
Demand &
Forecast
100
95
90
85
80
1 2 3 4 5 6 8 9 10 11
7 12
Period
Actual Moving Simple Exponential
Demand Average Smoothing
Thank You

OMT 8604 Logistics in Supply Chain Management: Master of Business Administration

Uploaded by

Copyright:

Available Formats

OMT 8604 Logistics in Supply Chain Management: Master of Business Administration

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

OMT 8604 Logistics in Supply Chain Management: Master of Business Administration

Uploaded by

Copyright:

Available Formats

OMT 8604 Logistics in Supply

D(2) E(1) D(3) F(2)

This figure suggest that SKU’s in a firm’s inventory

• Given the demand (Di) and value (vi) for

• Class A items should receive the most

Given a time series of past demand

• Similar procedures to estimate trends and

Where α is known as smoothing constant

You might also like