Forecasting Supply Chain Requirements

Forecasting Supply
Chain Requirements
Dr.Burcu Ozcam LOG301

The Importance of Forecasting
 Governments forecast unemployment, interest
rates, and expected revenues from income taxes
for policy purposes
 Marketing executives forecast demand, sales, and
consumer preferences for strategic planning
 College administrators forecast enrollments to plan
for facilities and for faculty recruitment
 Retail stores forecast demand to control inventory
levels, hire employees and provide training

What’s Forecasted in the
Supply Chain?
•Demand, sales or requirements
•Purchase prices
•Replenishment and delivery
lead times
CR (2004) Prentice Hall, Inc.

Some Forecasting Method Choices
•Historical projection
Moving average
Exponential smoothing
•Causal or associative
Regression analysis
•Qualitative
Surveys
Expert systems or rule-based
•Collaborative

Forecasting
 We focus on using historical data for forecasting
demand
 This should not diminish the importance of other

sources of information and common sense
 Information consists of
1. Historical data on our time series
2. Insight/knowledge and common sense
 Don’t confuse information with intuition
 Lets try a case study. Forecast a real time series from

scratch using intuition!
Monthly Sales and Forecast
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600 We’ll guess same as last month

400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
200 Monthly Sales

Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month

2000
1800
1600
1400
1200
Sales ($1000)
1000
We’ll guess same as last month
800 plus a little more for a possible trend
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
This is easy, who needs forecasting
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Continue with our successful method: guess
800 the same as last month plus a little more
for a possible trend
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Definitely looks like a trend
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Trend might be a tad steeper
800 than I thought
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Opps
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Momentary deviation,
800 trend will continue
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
See, I told you this was easy!
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
Trend will continue
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
Opps, another momentary fluctuation:
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400 Trend should continue

1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400 Oh oh!
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000 Sales has leveled off:

800
Lets average last few points
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000 Oh oh, maybe things are

800 going down hill
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
Let’s be conservative and

1000
Assume a negative trend
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
Thank goodness, we are

1000
still basically level
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
We’ll guess same as

1000 last month
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
This stuff is easy
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
We have for sure
1400
leveled off
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600 Big trouble!!!

Chief forecaster Smith and
1400
CEO Smothers fired!
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400 New chief forecaster points

1200
out the obvious trend
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
Remarkable turnaround in sales.
1400
New CEO Smithers given credit
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200 Still looks like a trend to me

Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200 Maybe not!

Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200 Level except for anomaly

Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200 Have things turned around?

Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
I’ll hedge my bets
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800 Things have turned around.

1600 Perhaps Smithers truly is a genius
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
Trend up!
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
Not bad!
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
Revise trend a tad
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800 Smithers makes cover of Fortune

1600
1400
1200
Sales ($1000)
1000
800
600
400
Smithers
200
Smothers Monthly Sales
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800 This is easy!!

1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
No big deal,
1600
trend continues
1400
1200
Sales ($1000)
1000
800
(in an unrelated matter
Smithers cashes out
600
stock options)
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800 Heads will surely

roll soon
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800 Let’s be cautiously

optimistic
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
Smithers called
1600 before board
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800 Perhaps we over

1600 reacted
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600 We will guess level

400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600 Back to normal!

400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800
1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
2000
1800 Smithers fired!

1600
1400
1200
Sales ($1000)
1000
800
600
400
Monthly Sales
200
Forecast
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Month
What have we learned?
 Our Actual sales appears to be a great leading indicator of our forecast
• It is supposed to work the other way around!!!!
 If we add up the (absolute value of) our forecast errors, we get 226.2
 If we had simply guessed “same as last month” we get 175.1
 Our intuition (ability to recognize a pattern) was poor given almost no

information or data. Never-the-less we saw patterns.
 For monthly data we can be tempted to “over think” forecasting.
 Now some additional information:

• Source of data is monthly sales of Australian Red Wine
• We also have a few years of data
M onthly S ales of A us tralian Red W ine
3500 clear seasonal behavior

clear upward trend
increase in amplitude
3000
2500
Sales($1000)
2000
1500
1000
500
0
1
8
15
22
29
36
43
85
50
57
64
71
78
92
99
106
113
120
127
134
141
M o n th
Value of Data
 Given data, we can forecast this series quite
accurately.
 This assumes stable behavior
 Recommend at least 4 - 5 seasons of data.
 Monthly demand thus prefers 4 to 5 years of data
 With 2 years of data, we are essentially forecasting on
the basis of two points if there is seasonal behavior

Laws of forecasting
1.We assume the future will behave like the past
• In the real world, the future often does not behave like the past.
1. Even given that the future behaves like the past, there is a limit to how accurate
forecasts can be (or nothing can be predicted with complete accuracy)
• The key issue is: How close will the forecast be to the actual value?
• It is crucial to attempt to quantify the expected accuracy of a forecast
1. The further into the future you attempt to forecast, the greater will be the forecast error.
• Major decisions are often based on long term forecasts. e.g. building a new plant
• Considering risk is even more important in these cases
1. Decisions will be based on the forecast (Otherwise there is no need to forecast!)

• That is: forecasts have inherent error, thus the decisions based on forecasts have
inherent risk

P ast D ata and Future Fore casts
20
15 Now
10
5
Demand
-5
Past Data Future forecast
-10
-15
-20
1
5
9
13
21
25
29
33
41
57
65
69
73
17
37
45
49
53
61
P e rio d
Fore casts with 50% P re diction Inte rv als
20
15
10
5
Demand
-5
-10
-15
-20
1
5
9
13
21
25
29
33
41
57
65
69
73
17
37
45
49
53
61
P e rio d
Fore casts with 95% P re diction Inte rv als
20
15
10
5
Demand
-5
-10
-15
-20
1
5
9
13
21
25
29
33
41
57
65
69
73
17
37
45
49
53
61
P e rio d
Time-Series Data
 Numerical data obtained at regular time
intervals
 The time intervals can be annually, quarterly,
daily, hourly, etc.
 Example:
Year: 1999 2000 2001 2002 2003
Sales: 75.3 74.2 78.5 79.7 80.2

Time Series Plot
A time-series plot is a two-dimensional
plot of time series data
 the vertical axis U.S. Inflation Rate

measures the variable
16.00
of interest 14.00
12.00
Inflation Rate (%)
10.00
 the horizontal axis 8.00
corresponds to the 6.00
4.00
time periods 2.00
0.00
1975
1977
1979
1981
1989
1993
1995
1997
1999
2001
1983
1985
1987
1991
Year
Time-Series Components
Time-Series
Trend Seasonal Cyclical Random

Component Component Component Component

Trend Component
 Long-run increase or decrease over time (overall
upward or downward movement)
 Data taken over a long period of time
Sales rd t re nd
U pwa

Time
Trend Component (continued)
 Trend can be upward or downward
 Trend can be linear or non-linear
Sales Sales
Time Time
Downward linear trend Upward nonlinear trend

Seasonal Component
 Short-term regular wave-like patterns
 Observed within 1 year
 Often monthly or quarterly
Sales
Summer
Winter
Spring Fall
Time (Quarterly)
Cyclical Component
 Long-term wave-like patterns
 Regularly occur but may vary in length
 Often measured peak to peak or trough to trough
1 Cycle
Sales
Dr.Burcu Ozcam LOG301 Year

Random Component
 Unpredictable, random, “residual” fluctuations
 Due to random variations of

• Nature
• Accidents or unusual events
 “Noise” in the time series

Trend-Based Forecasting
 Estimate a trend line using regression analysis
Time  Use time (t) as the

Year Period Sales independent variable:
(t) (y)
1999 1 20
2000 2 40 ŷ = b0 + b1t
2001 3 30
2002 4 50
2003 5 70
2004 6 65

 The linear trend model is:
Time
Year Period Sales
ŷ = 12.333 + 9.5714 t
(t) (y)
1999 1 20 Sales trend
2000 2 40 80
70
2001 3 30 60
2002 4 50 50
sales
40
2003 5 70 30
20
2004 6 65 10
0
0 1 2 3 4 5 6
 Forecast for time period 7:
Time
Year Period (t) Sales ŷ = 12.333 + 9.5714 (7)
(y)
1999 1 20 = 79.33Sales
2000 2 40
80
2001 3 30 70
2002 4 50 60
2003 5 70 50
sales
40
2004 6 65 30
2005 7 ?? 20
10
0
0 1 2 3 4 5 6 7
Comparing Forecast Values
to Actual Data
 The forecast error or residual is the difference
between the actual value in time t and the forecast
value in time t:
 Error in time t:
e t = y t − Ft

Two common Measures of Fit
 Measures of fit are used to gauge how well the
forecasts match the actual values
MSE (mean squared error)

• Average squared difference between yt and Ft
MAD (mean absolute deviation)

• Average absolute value of difference between yt and Ft
• Less sensitive to extreme values

MSE vs. MAD
Mean Square Error Mean Absolute Deviation
MSE =
∑ (y t − Ft ) 2
MAD =
∑ |y t − Ft |
n n
where:
yt = Actual value at time t
Ft = Predicted value at time t
n = Number of time periods
Moving Averages
 Used for smoothing
 Series of arithmetic means over time
 Result dependent upon choice of L (length of
period for computing means)
 To smooth out seasonal variation, L should
be equal to the number of seasons
• For quarterly data, L = 4
• For monthly data, L = 12

Moving Averages (continued)
 Example: Four-quarter moving average
• First average:
Q1 + Q2 + Q3 + Q4
Moving average 1 =
4
• Second average:
Q2 + Q3 + Q4 + Q5
Moving average 2 =
4
• etc…

Seasonal Data
Quarter Sales
1 23 Quarterly Sales
2 40
60
3 25
4 27
50
…
40
5 32
Sales
6 48 30
7 33 20
8 37 10 …
9 37 0
10 50 1 2 3 4 5 6 7 8 9 10 11
11 40 Quarter
etc… etc…

Calculating Moving Averages
Average 4-Quarter
Period Moving
Quarter Sales Average
1+ 2 + 3 + 4
2.5 28.75 2.5 =
1 23 4
3.5 31.00
2 40 23 + 40 + 25 + 27
4.5 33.00 28.75 =
3 25 4
5.5 35.00
4 27 6.5 37.50
5 32 7.5 38.75
etc…
6 48 8.5 39.25
7 33 9.5 41.00
8 37
 Each moving average is for a
9 37 consecutive block of 4 quarters
10
Dr.Burcu Ozcam 50 LOG301
Single Exponential Smoothing
 A weighted moving average
• Weights decline exponentially
• Most recent observation weighted most
 Used for smoothing and short term forecasting

Single Exponential Smoothing(continued)
 The weighting factor is α
• Subjectively chosen
• Range from 0 to 1
• Smaller α gives more smoothing, larger α
gives less smoothing
 The weight is:

• Close to 0 for smoothing out unwanted cyclical
and irregular components
• Close to 1 for forecasting

Exponential Smoothing Model
 Single exponential smoothing model
Ft +1 = Ft + α( y t − Ft )
or:
Ft +1 = αy t + (1 − α )Ft
where:
Ft+1 = forecast value for period t + 1
yt = actual value for period t
Ft = forecast value for period t
α = alpha (smoothing constant)
Exponential Smoothing Example
 Suppose we use weight α = .2
Quarter (t) Sales Forecast Forecast for next period (Ft+1 )
(yt) from prior
period
F1 = y1 since
1 23 NA 23 no prior
2 40 23 (.2)(40)+(.8)(23)=26.4 information
3 25 26.4 (.2)(25)+(.8)(26.4)=26.12 exists
4 27 26.12 (.2)(27)+(.8)(26.12)=26.296
5 32 26.296 (.2)(32)+(.8)(26.296)=27.437
6 48 27.437 (.2)(48)+(.8)(27.437)=31.549 Ft +1
7 33 31.549 (.2)(48)+(.8)(31.549)=31.840 = αy t + (1 − α)Ft
8 37 31.840 (.2)(33)+(.8)(31.840)=32.872
9 37 32.872 (.2)(37)+(.8)(32.872)=33.697
10 50 33.697 (.2)(50)+(.8)(33.697)=36.958
etc… etc… etc… etc…

Sales vs. Smoothed Sales
 Seasonal
fluctuations have
been smoothed 60
 NOTE: the 50
smoothed value in
40
this case is Sales
generally a little low, 30
since the trend is 20
upward sloping and
the weighting factor 10
is only .2 0
1 2 3 4 5 6 7 8 9 10
Quarter
Sales Smoothed

Double Exponential Smoothing
 Double exponential smoothing is sometimes called
exponential smoothing with trend
 If trend exists, single exponential smoothing may need
adjustment
 Add a second smoothing constant to account for trend

Model
C t = αy t + (1 − α )(C t −1 + Tt −1 )
Tt = β(C t − C t −1 ) + (1 − β)Tt −1
Ft +1 = C t + Tt
where:
yt = actual value in time t
α = constant-process smoothing constant
β = trend-smoothing constant
Ct = smoothed constant-process value for period t
Tt = smoothed trend value for period t
Ft+1 = forecast value for period t + 1
Dr.Burcu Ozcam t = current time period LOG301
 Double exponential smoothing is generally

done by computer
 Use larger smoothing constants α and β when
less smoothing is desired
 Use smaller smoothing constants α and β
when more smoothing is desired

Exponential Smoothing in Excel
 Use tools / data analysis /
exponential smoothing
• The “damping factor” is (1 - α )

Types of Regression Models
Positive Linear Relationship Relationship NOT Linear
Negative Linear Relationship No Relationship

Simple Linear Regression Model
 Only one independent variable, x
 Relationship between x and y is
described by a linear function
 Changes in y are assumed to be caused
by changes in x

Linear Regression
The regression model:
Slope Random
y intercept Independent Error
Coefficient
Variable term, or
Dependent residual
y = b 0 + b1x + ε
Variable
Linear component Random Error

component

Linear Regression
(continued)
y y = b 0 + b1x + ε
Observed Value
of y for xi
εi Slope = b1
Predicted Value Random Error
of y for xi
for this x value
Intercept = b0
xi x
Least Squares Criterion
 b0 and b1 are obtained by finding the values

of b0 and b1 that minimize the sum of the
squared residuals
∑e 2
= ∑ (y −ŷ) 2
= ∑ (y − (b 0 + b1x))
2

The Least Squares Equation
 The formulas for b1 and b0 are:
b1 =
∑ ( x − x )( y − y )
∑ (x − x) 2
algebraic equivalent:
and
∑ xy − ∑ x∑ y
b1 = n b0 = y − b1 x
∑ x 2
−
( ∑ x ) 2
n
Interpretation of the
Slope and the Intercept
 b0 is the estimated average value of y when the
value of x is zero
 b1 is the estimated change in the average value

of y as a result of a one-unit change in x

Finding the Least Squares Equation
 The coefficients b0 and b1 will usually be found
using computer software, such as Excel or
Minitab
 Other regression measures will also be computed

as part of computer-based regression analysis

Simple Linear Regression Example
 A real estate agent wishes to examine the

relationship between the selling price of a home
and its size (measured in square feet)
 A random sample of 10 houses is selected

• Dependent variable (y) = house price in
$1000s
• Independent variable (x) = square feet

Sample Data for House Price Model
House Price in $1000s Square Feet
(y) (x)
245 1400
312 1600
279 1700
308 1875
199 1100
219 1550
405 2350
324 2450
319 1425
255 1700
Regression Using Excel
 Tools / Data Analysis / Regression

Excel Output
Regression Statistics
Multiple R 0.76211
The regression equation is:
R Square 0.58082
Adjusted R Square 0.52842 house price = 98.24833 + 0.10977 (square feet)
Standard Error 41.33032
Observations 10
ANOVA df SS MS F Significance F
Regression 1 18934.9348 18934.9348 11.0848 0.01039

Residual 8 13665.5652 1708.1957
Total 9 32600.5000
Coefficients Standard Error t Stat P-value Lower 95% Upper 95%

Intercept 98.24833 58.03348 1.69296 0.12892 -35.57720 232.07386
Square Feet 0.10977 0.03297 3.32938 0.01039 0.03374 0.18580

Graphical Presentation
 House price model: scatter plot and regression line
450
400
350 Slope
House Price ($1000s)
300 = 0.10977
250
200
150
100
Intercept 50
= 98.248 0
0 500 1000 1500 2000 2500 3000
Square Feet
house price = 98.24833 + 0.10977 (square feet)
Actions When Forecasting is
Not Appropriate
• Seek information directly from customers
•Collaborate with other channel members
• Apply forecasting methods with caution (may work
where forecast accuracy is not critical)
• Delay supply response until demand
becomes clear
• Shift demand to other periods for better
supply response
• Develop quick response and flexible supply
systems
Collaborative Forecasting
• Demand is lumpy or highly uncertain
• Involves multiple participants each with
a unique perspective—“two heads are
better than one”
• Goal is to reduce forecast error
• The forecasting process is inherently
unstable

Collaborative Forecasting:
Key Steps
• Establish a process champion
• Identify the needed Information and collection processes
• Establish methods for processing information from multiple
sources and the weights assigned to multiple forecasts
• Create methods for translating forecast into form needed by
each party
• Establish process for revising and updating forecast in real
time
• Create methods for appraising the forecast
• Show that the benefits of collaborative forecasting are obvious
and real

Managing Highly
Uncertain Demand
•Delay forecasting as long as possible
•Prioritize supply by product’s degree of uncertainty
(supply to the more certain products first)
•Apply the principle of postponement to the most
uncertain products (delay committing to a final product
form until an order is received)
•Create flexible supply to changing demand (alter
capacity and output rates through subcontracting,
computer technology, multi-purpose processes, etc.)
•Be able to respond quickly to uncertain demand levels
 Ch 2, Problems 1,4,8,9,10
 Due 8-12 Oct.

Forecasting Supply Chain Requirements

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Forecasting Supply Chain Requirements

Hochgeladen von

Copyright:

Verfügbare Formate

Forecasting Supply

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

CR (2004) Prentice Hall, Inc.

Dr.Burcu Ozcam LOG301

 This should not diminish the importance of other

 Don’t confuse information with intuition

 Lets try a case study. Forecast a real time series from

600 We’ll guess same as last month

200 Monthly Sales

Dr.Burcu Ozcam LOG301

1400 Trend should continue

1000 Sales has leveled off:

1000 Oh oh, maybe things are

Let’s be conservative and

Thank goodness, we are

We’ll guess same as

1600 Big trouble!!!

1400 New chief forecaster points

1200 Still looks like a trend to me

1200 Maybe not!

1200 Level except for anomaly

1200 Have things turned around?

1800 Things have turned around.

1800 Smithers makes cover of Fortune

1800 This is easy!!

800 Heads will surely

800 Let’s be cautiously

1800 Perhaps we over

600 We will guess level

600 Back to normal!

1800 Smithers fired!

 If we had simply guessed “same as last month” we get 175.1

 Our intuition (ability to recognize a pattern) was poor given almost no

 For monthly data we can be tempted to “over think” forecasting.

 Now some additional information:

3500 clear seasonal behavior

Dr.Burcu Ozcam LOG301

1. Decisions will be based on the forecast (Otherwise there is no need to forecast!)

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

 the vertical axis U.S. Inflation Rate

Trend Seasonal Cyclical Random

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301 Year

 Due to random variations of

 “Noise” in the time series

Dr.Burcu Ozcam LOG301

Time  Use time (t) as the

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

MSE (mean squared error)

MAD (mean absolute deviation)

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301

 The weight is:

Dr.Burcu Ozcam LOG301

Dr.Burcu Ozcam LOG301