Operations Management

Forecasting
 Why Forecasting?

 Decision making is often complicated due to

uncertainty
 Since Each decision involves considerable cash flow,
time & other resource, good estimate is a crucial
requirement
 Forecast are estimate of timing & magnitude of the
occurrence of future events.
 Primary Function is to Predict the Future using (time
series related or other) data we have in hand
 Forecasting involves a study of the present & past
data with a view to estimate the future.
Why are we interested?

 Planning is fundamental to Management

 Affects the decisions we make today
 The better the Management is able to forecast
the future the better will it be prepared to
face the future.
Where is forecasting used in
POM

 Forecast dynamic & complex environment

 Forecast short term fluctuations in production
 forecast availability/need for manpower
 Better materials management
 Basis for planning & scheduling
 Strategic decision.
 What Makes a Good Forecast?

 It should be as accurate as possible

 It should be reliable
 It should be in meaningful units
 It should be presented in writing
 The method should be easy to use and
understand in most cases.
Forecast Horizons in Operation
Planning – Figure 2.1
 Types of Forecasts by Time Horizon

 Short-range forecast
 Up to 1 year; usually less than 3 months

 Employed to fine tune an existing plan

 Act as an input to tactical decisions

 Degree of uncertainty is low

 Job scheduling, worker assignments

 Medium-range forecast
 Planning horizon is usually 12 to 18 months

 Aggregation of data is done considering cyclical &

seasonal patterns
  Nature of decision will be tactical as well as strategic
 E.g. annual production planning, capacity augmentation,

 Long-range forecast
 Involves purely strategic decision for time period of about

5 to 10 years
 It involves subjective knowledge from the experts

 Level of uncertainty is high

 E.g. New product planning, facility location

 Forecasts are the basis for budgeting, planning capacity,
sales, production and inventory, personnel, purchasing etc

Forecasts affects decisions in all the departments

in an organization
 Accounting – new product estimated cost, profit projections
 Finance – replacement of equipment, amount of
funding/borrowing needs
 Human Resources – hiring activities, layoff planning
 Marketing – pricing and promotions etc.
 MIS – new/revised information systems
 Operations – schedules, capacity planning, work assignments,
inventory planning, make-or-buy decisions, outsourcing etc.
 Product/service design – timeline to design a new product etc.
 Sources of data

 Sales force estimate

 Point of sales (pos)data systems
 Forecasts from supply chain partners
 Trade / industry association journals
 B2B portals / market places
 Economic surveys & indicators
 Subjective knowledge
Qualitative Methods Quantitative
Methods
 Subjective approach
 Objective in nature
 Used when situation is  Used when situation is
vague & little data exist ‘stable’ & historical data
 New products exist
 New technology  Existing products

 Current technology
 Involves intuition,
experience  Involves mathematical
 e.g., forecasting sales techniques
on Internet  e.g., forecasting sales

of color televisions
 Quantitative Methods

 Extrapolative methods
 Make use of past data
 E.g. moving avg method, weighted moving average,
exponential method
 Causal models
 Analyze data from the point view of cause & effect
relationship.
 E.g. demand for new houses based on other factors
Subjective Forecasting Methods

 Sales Force Composites

 Aggregation of sales personnel estimates
 Customer Surveys
 Jury of Executive Opinion
 The Delphi Method
 Jury of Executive
Opinion
 Involves small group of high-level
managers
 Group estimates demand by working

together
 Combines managerial experience with
statistical models
 Relatively quick
 ‘Group-think’
disadvantage

© 1995 Corel Corp.

Sales Force Composite

Sales
 Each salesperson
projects his or her
sales
 Combined at district
& national levels
 Sales reps know
customers’ wants
 Tends to be overly
optimistic
 Delphi Method

 The Delphi Method seeks to achieve a consensus among group

members through a series of questionnaires. The questionnaires are
answered anonymously and individually by each member of the group.
 The answers are summarized and sent back to the group members
along with the next questionnaire. This process is repeated until a group
consensus is reached.
 This usually only takes two iterations, but can sometimes take as many
as six rounds before a consensus is reached.
 When Should You Use the
Delphi Method ?
 When the number of participants make face-to-face
meetings impossible.
 When the group members are scattered across a large
geographic area.
 When face-to-face meetings are being dominated by
one individual.
 When accurate information is not available or
expensive to collect.
Consumer Market Survey

How many hours

 Ask customers
will you use the
about purchasing
Internet next week?
plans
 What consumers
say, and what
they actually do
are often
different
 Sometimes
difficult to ©
answer 1995
Corel
 Demand Behavior

 Trend
 a gradual, long-term up or down movement of
demand
 Random variations
 movements in demand that do not follow a pattern
 Cycle
 an up-and-down repetitive movement in demand
(over lengthy time span i.e., more than a year)
 Seasonal pattern
 an up-and-down repetitive movement in demand
occurring periodically
 Forms of Forecast Movement

Demand
Demand

Random
movement

Time Time( yrs)

(a) Trend (b) Cycle

Demand
Demand

Time (mths) Time

(c) Seasonal pattern (d) Trend with seasonal pattern
 Time Series
 Assume that what has occurred in the past will
continue to occur in the future
 Relate the forecast to only one factor - time
 Include
 moving average
 exponential smoothing
 linear trend line
 Moving Average

 Naive forecast
 demand in the current period is used as next
period’s forecast
 Simple moving average
 stable demand with no pronounced
behavioral patterns
 Weighted moving average
 weights are assigned to most recent data
Moving Average:
Naïve Approach
ORDERS
MONTH PER MONTH FORECAST

Jan 120 -
Feb 90 120
Mar 100 90
Apr 75 100
May 110 75
June 50 110
July 75 50
Aug 130 75
Sept 110 130
Oct 90 110
Nov - 90
Simple Moving Average

Σ
i = 1 Di
MAn =
n
where

n = number of periods
in the moving
average
Di = demand in period i
 3-month Simple Moving Average

3
ORDERS MOVING
AVERAGE
Σ D
i=1 i
MONTH PER MA3 =
Jan 120 – 3
MONTH
–
Feb 90 – 90 + 110 + 130
103.3 = 3
Mar 100 88.3
95.0
Apr 75 78.3 = 110 orders
78.3 for Nov
May 110 85.0
105.0
June 50 110.0

July 75
 5-month Simple Moving Average

ORDERS MOVING
5
AVERAGE
MONTH
Jan
PER
120 –
Σ D
i=1 i
MONTH MA5 =
–
5
Feb 90 –
–
90 + 110 + 130+75+50
Mar 100 – =
99.0
5
Apr 75 85.0
82.0 = 91 orders
May 110 88.0 for Nov
95.0
June 50 91.0

July 75
 Smoothing Effects
150 –

125 – 5-month

100 –

75 –
Orders

50 –

3-month
25 –
Actual
0–

| | | | | | | | | | |
Jan Feb Mar Apr May June July Aug Sept Oct Nov
Month
 Weighted Moving Average

 Adjusts WMAn =Σ Wi Di
i=1 i=1
moving
average where
method to Wi = the weight for period i,
more closely between 0 and 100
percent
reflect data
fluctuations
Σ Wi = 1.00
Weighted Moving Average Example

MONTH WEIGHT DATA

August 17% 130
September 33% 110
October 50% 90
3
November Forecast WMA3 =Σ
i = 1 Wi Di

= (0.50)(90) + (0.33)(110) + (0.17)(130)

= 103.4 orders
 Exponential Smoothing

 Averaging method
 Weights most recent data more strongly
 Reacts more to recent changes
 Widely used, accurate method
 Exponential Smoothing (cont.)

Ft +1 = α Dt + (1 - α )Ft
where:
Ft +1 = forecast for next period
Dt = actual demand for present period
Ft = previously determined forecast
for present period
α = weighting factor, smoothing constant
 Effect of Smoothing Constant

0.0 ≤ α ≤ 1.0
If α = 0.20, then Ft +1 = 0.20 Dt + 0.80 Ft

If α = 0, then Ft +1 = 0 Dt + 1 Ft 0 = Ft
Forecast does not reflect recent data

If α = 1, then Ft +1 = 1 Dt + 0 Ft = Dt
Forecast based only on most recent data
 Exponential Smoothing (α=0.30)

PERIOD MONTH F2 = α D1 + (1 - α )F1

DEMAND
= (0.30)(37) + (0.70)(37)
1 Jan 37 = 37

2 Feb 40 F3 = α D2 + (1 - α )F2
= (0.30)(40) + (0.70)(37)
3 Mar 41
= 37.9
4 Apr 37
F13 = α D12 + (1 - α )F12

5 May 45 = (0.30)(54) + (0.70)(50.84)

= 51.79
6 Jun 50

7 Jul 43
Exponential Smoothing
(cont.)
FORECAST, Ft +1
PERIOD MONTH DEMAND (α = 0.3) (α = 0.5)
1 Jan 37 – –
2 Feb 40 37.00 37.00
3 Mar 41 37.90 38.50
4 Apr 37 38.83 39.75
5 May 45 38.28 38.37
6 Jun 50 40.29 41.68
7 Jul 43 43.20 45.84
8 Aug 47 43.14 44.42
9 Sep 56 44.30 45.71
10 Oct 52 47.81 50.85
11 Nov 55 49.06 51.42
12 Dec 54 50.84 53.21
13 Jan – 51.79 53.61
 Exponential Smoothing (cont.)
70 –

60 – Actual α = 0.50
50 –

40 –
Orders

30 – α = 0.30
20 –

10 –

0–

| | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Month
Adjusted Exponential Smoothing

AFt +1 = Ft +1 + Tt +1
where
T = an exponentially smoothed trend factor

Tt +1 = β (Ft +1 - Ft) + (1 - β ) Tt
where
Tt = the last period trend factor
β = a smoothing constant for trend
 Adjusted Exponential
Smoothing (β=0.30)
T3 = β (F3 - F2) + (1 - β ) T2
PERIOD MONTH
= (0.30)(38.5 - 37.0) + (0.70)(0)
DEMAND
= 0.45
1 Jan 37
AF3 = F3 + T3 = 38.5 + 0.45
2 Feb 40 = 38.95

3 Mar 41 T13 = β (F13 - F12 ) + (1 - β ) T12

= (0.30)(53.61 - 53.21) + (0.70)
4 Apr 37 (1.77)
= 1.36
5 May 45
AF13 = F13 + T13 = 53.61 + 1.36 = 54.96
6 Jun 50
 Adjusted Exponential Smoothing:
Example
FORECAST TREND ADJUSTED
PERIOD MONTH DEMAND Ft +1 Tt +1 FORECAST AFt +1

1 Jan 37 37.00 – –
2 Feb 40 37.00 0.00 37.00
3 Mar 41 38.50 0.45 38.95
4 Apr 37 39.75 0.69 40.44
5 May 45 38.37 0.07 38.44
6 Jun 50 38.37 0.07 38.44
7 Jul 43 45.84 1.97 47.82
8 Aug 47 44.42 0.95 45.37
9 Sep 56 45.71 1.05 46.76
10 Oct 52 50.85 2.28 58.13
11 Nov 55 51.42 1.76 53.19
12 Dec 54 53.21 1.77 54.98
13 Jan – 53.61 1.36 54.96
 Adjusted Exponential Smoothing
Forecasts
70 –

60 – Adjusted forecast (β = 0.30)

Actual
50 –

40 –
Demand

30 –
Forecast (α = 0.50)
20 –

10 –

0–

| | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Period
 Trend projections

 This technique fits a trend line to a series

of historical data points & then projects
the line into the future for medium to long
range forecast.
 Linear Trend Line

Σ xy -
y = a + bx b =
nxy
a = y - Σb x x2 -
where
a = intercept nx2
where
b = slope of the line n = number of periods
x = usually time period
Σ x
(independent variable) x = = mean of the x values
n
y = forecast for
demand for period x Σ y
y = n = mean of the y values
(dependent variable)
Least Squares Example
x(PERIOD) y(DEMAND) xy x2
1 73 37 1
2 40 80 4
3 41 123 9
4 37 148 16
5 45 225 25
6 50 300 36
7 43 301 49
8 47 376 64
9 56 504 81
10 52 520 100
11 55 605 121
12 54 648 144
78 557 3867 650
 Least Squares Example
(cont.)
78
x = = 6.5
12
557
y = = 46.42
12
∑xy - nxy 3867 - (12)(6.5)(46.42)
b = = =1.72
∑x - nx
2 2
650 - 12(6.5) 2

a = y - bx
= 46.42 - (1.72)(6.5) = 35.2
 Linear trend line y = 35.2 + 1.72x
Forecast for period 13 y = 35.2 + 1.72(13) = 57.56 units

70 –

60 –
Actual
50 –
Demand

40 –

30 – Linear trend line

20 –

10 –

0–
| | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Period
 Seasonal Adjustments

 Repetitive increase/ decrease in demand

 Use seasonal factor to adjust forecast

Di
Seasonal factor = Si =
∑D
Month 2005 2006 2007 Avg(2005- Avg Seasonal
07) demand month index
demand
Jan 80 85 105 90 94 .957(=90/94
Feb 70 85 85 80 94 .851
Mar 80 93 82 85 94 .904
Apr 90 95 115 100 94 1.064
May 113 125 131 123 94 1.309
June 110 115 120 115 94 1.223
July 100 102 113 105 94 1.117
Aug 88 102 110 100 94 1.064
Sep 85 90 95 90 94 .957
Oct 77 78 85 80 94 .851
Nov 75 82 83 80 94 .851
Dec 82 78 80 80 94 .851
1128/12=94
 Forecast Accuracy

 Forecast error
 difference between forecast and actual demand
 Forecast error = Actual demand – Forecast value

 Some Popular measures are

 SFE (Sum of forecast error)
 MAD (mean absolute deviation)
 MAPD (mean absolute percent deviation)
 MSE ( mean squared error)
 Sum of forecast error(SFE)

 Sum of errors during the period of consideration.

SFE= ∑ ℮t
 Forecast error ℮t= Dt – Ft
 if ℮t is positive it indicates underestimation of
demand & vice versa.
 Mean Absolute Deviation
(MAD)
To overcome limitation of SFE one could take absolute values of
Error & avg it. Σ | D -F|
t t
MAD = n
where
t = period number
Dt = demand in period t
Ft = forecast for period t
n = total number of periods
  = absolute value
 MAD Example
PERIOD DEMAND, Dt Ft (α =0.3) (Dt - Ft) | Dt - F t |
1 37 37.00 – –
2 40 37.00 3.00 3.00
3 Σ | D41 -
t Ft | 37.90 3.10 3.10
MAD
4 = 37
n
38.83 -1.83 1.83
5 45 38.28 6.72 6.72
6 53.39
50 40.29 9.69 9.69
7 = 43 43.20 -0.20 0.20
8 11
47 43.14 3.86 3.86
9 56 44.30 11.70 11.70
10 = 4.8552 47.81 4.19 4.19
11 55 49.06 5.94 5.94
12 54 50.84 3.15 3.15
557 49.31 53.39
 Other Accuracy Measures

Mean absolute percent deviation (MAPD)

∑|Dt - Ft|
MAPD =
∑Dt
Expressing error in relative term rather than absolute terms

In Situations which demand low tolerance amplifying of

forecast error is desirable.
Mean Squared error
∑(forecast error)²
MSE =
n
 Forecast Control
 Tracking signal
 Is a measurement of how well a forecast is

predicting actual values

 monitors the forecast to see if it is biased high or low

Tracking signal = ∑(Dt - Ft) = RSFE

MAD MAD
 +ve signals indicate that demand is greater than
forecast & vice versa.
 Control limits of 2 to 5 MADs are used most
frequently
 Tracking Signal Values
DEMAND FORECAST, ERROR ∑E = TRACKING
PERIOD Dt Ft Dt - Ft ∑ (Dt - Ft) MAD SIGNAL

1 37 37.00 – – – –
2 40 37.00 3.00 3.00 3.00 1.00
3 41 37.90 3.10 6.10 3.05 2.00
4 37 38.83 -1.83 4.27 2.64 1.62
5 45 38.28
Tracking 6.72 for period
signal 10.99 3 3.66 3.00
6 50 40.29 9.69 20.68 4.87 4.25
7 43 43.20 -0.20 20.48 4.09 5.01
6.10
8 47 TS3 = 3.86 =24.34
43.14 2.00 4.06 6.00
9 56 44.30 3.05 36.04
11.70 5.01 7.19
10 52 47.81 4.19 40.23 4.92 8.18
11 55 49.06 5.94 46.17 5.02 9.20
12 54 50.84 3.15 49.32 4.85 10.17
 Tracking Signal Plot
3σ –

2σ –
Tracking signal (MAD)

Exponential smoothing (α = 0.30)

1σ –

0σ –

-1σ –

-2σ –

-3σ – Linear trend line

| | | | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10 11 12
Period
 Statistical Control Charts

∑(Dt - Ft)2
σ =
n-1

 Using σ we can calculate statistical

control limits for the forecast error
 Control limits are typically set at ± 3σ
 Statistical Control Charts
18.39 –
UCL = +3σ
12.24 –

6.12 –

0–
Errors

-6.12 –

-12.24 –

-18.39 –
LCL = -3σ

| | | | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10 11 12
Period
 Regression Methods

 Linear regression
 a mathematical technique that relates a
dependent variable to an independent
variable in the form of a linear equation
 Correlation
 a measure of the strength of the relationship
between independent and dependent
variables
 Linear Regression

y = a + bx a = y-bx
Σ xy -
b =
nxy
where Σ x2 -
a = intercept
nx2
b = slope of the line
Σ x
x = = mean of the x data
n
Σ y
y = n = mean of the y data
 Linear Regression Example
x y
(WINS) (ATTENDANCE) xy x2
4 36.3 145.2 16
6 40.1 240.6 36
6 41.2 247.2 36
8 53.0 424.0 64
6 44.0 264.0 36
7 45.6 319.2 49
5 39.0 195.0 25
7 47.5 332.5 49
49 346.7 2167.7 311
Linear Regression Example (cont.)
49
x= = 6.125
8
346.9
y= = 43.36
8

∑xy - nxy2
b=
∑x2 - nx2
(2,167.7) - (8)(6.125)(43.36)
=
(311) - (8)(6.125)2
= 4.06

a = y - bx
= 43.36 - (4.06)(6.125)
= 18.46
 Linear Regression Example (cont.)
Regression equation Attendance forecast for 7 wins
y = 18.46 + 4.06x y = 18.46 + 4.06(7)
60,000 – = 46.88, or 46,880

50,000 –

40,000 –
Attendance, y

30,000 –

20,000 –
Linear regression line,
10,000 –
y = 18.46 + 4.06x

| | | | | | | | | | |
0 1 2 3 4 5 6 7 8 9 10
Wins, x
 Correlation and Coefficient of
Determination
 Correlation, r
 Measure of strength of relationship
 Varies between -1.00 and +1.00
 Coefficient of determination, r2
 Percentage of variation in dependent
variable resulting from changes in the
independent variable
 Computing Correlation
n∑ xy - ∑ x∑ y
r=
[n∑ x2 - (∑ x)2] [n∑ y2 - (∑ y)2]

(8)(2,167.7) - (49)(346.9)
r=
[(8)(311) - (49)2 ] [(8)(15,224.7) - (346.9)2]

r = 0.947

Coefficient of determination
r2 = (0.947)2 = 0.897
Multiple Regression
Study the relationship of demand to two or
more independent variables

y = β 0 + β 1x1 + β 2x2 … + β kxk

where
β 0 = the intercept
β 1, … , β k = parameters for the
independent variables
x1, … , xk = independent variables