OPERATIONS MANAGEMENT- II

(ORM2BJ21-3)

Alok Raj ([email protected])

Associate Professor, PODS area
Office: Room number 14, 2nd Floor, Library Building
Course details
Course Instructor(s)
 Alok Raj ([email protected])

 Abhishek Chakraborty ([email protected])

Textbook
Operations Management by Jay H. Heizer

Course Pages
Link: https://classroom.google.com/c/NjQ2NzA0NDM3OTIx
Access Code: yt5e57s
Class Rules
Evaluation
 Quiz, Midterm

 Each carrying a weightage of 25%

 Final grading will be done after adding all the marks obtained across all evaluation
components to a total of 100
Advanced Electric Cycle
Session Topics

 Forecasting

 Aggregate Requirement Planning

 Material Requirement Planning

 Scheduling

 Waiting Line Management

 Forecasting

Alok Raj
PODS Area
Office: Room No 14, 2nd Floor,
Library Building, Tel. 3439
Email: [email protected]
Objectives
 Understand the fundamental principles of forecasting.

 Learn various quantitative forecasting methods.

 Develop skills to select and apply appropriate forecasting

techniques.

 Analyze and interpret forecasting results.

 Case—Forecasting Beer Demand.

Forecasting

Products/
Services
Demand

Uncertainty
Good
Forecasting model

 Forecasts help in reducing uncertainties from a supply chain. A good forecast

helps in planning activities in a supply chain such as capacity investments,
production plans, delivery plans and schedules, ordering cycles and inventory
planning.
 Need of Forecasting because of better
 Planning
 Capacity Investment
 Delivery Plans
Which type of product category forecasting is more important ?

Push vs Pull system

Make-to-stock- Push- More relevance of

Forecasting?

Material Material

Make-to-Order- Pull
Types of Demand
 Independent demand (finished product)
 Dependent demand (component parts or subassemblies)

https://www1.grc.nasa.gov/wp-content/uploads/NASA-Glenn-Airplane-Parts-Image-2.jpg
What is the way for forecasting
First data: Diaper data sales
Year Actual
Demand
(×100000)
2014 25

2015 26

2016 24

2017 28

Actual Demand (×100000)

2018 26
30
28
2019 27
26
24
2020 26
22
20
2021 28 18
16
2022 25 14
12
2023 ? 10
2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
Second data: Passenger data sales

Passenger
Period Year Vehicle
Sales
2004 1061572
Passenger Vehicle Sales
1
2 2005 1143076 5000000
3 2006 1379979 4500000
4 2007 1549882
4000000
5 2008 1552703
6 2009 1951333 3500000
7 2010 2501542 3000000
8 2011 2629839
2500000
9 2012 2665015
10 2013 2503509 2000000
11 2014 2601111 1500000
12 2015 2789678
1000000
13 2016 3047582
14 2017 3288581 500000
15 2018 3377389 0
2019 2773575 2000 2005 2010 2015 2020 2025
16
17 2020 3062280
18 2021 3650698
19 2022 4578639
20 2023 ?
 Third data: Electric Sales data
Year E-2 Wheelers E-3 Wheelers E-4 Wheelers E-Buses Grand Total
17-18 2005 91970 2242 35 96252 EV Grand Total Sales
18-19 28007 116031 2407 75 146520 1400000
19-20 26834 143051 2404 369 172658
20-21 44803 90898 5201 373 141275 1200000
21-22 252641 172543 19782 1198 446164 1000000
22-23 728054 401882 48105 1917 1179958
800000
Grand Total 1082344 1016375 80141 3967 2182827
600000
400000
200000
0
17-18 18-19 19-20 20-21 21-22 22-23
Electric Vehicle Sale
1400000

1200000

1000000

800000

600000

400000

200000

0
17-18 18-19 19-20 20-21 21-22 22-23

E-2 Wheelers E-3 Wheelers E-4 Wheelers E-Buses Grand Total

Fourth data: Air passenger movements data
Air passenger
Sr
Year Month movements (in
no
Lakhs)
13 Jan-16 242
14 Feb-16 233 Air passenger movements (in Lakhs)
15 Mar-16 267
16 Apr-16 269 450
17 May-16 270
18 Jun-16 315 400
2016
19 Jul-16 364
20 Aug-16 347
350
21 Sep-16 312
22 Oct-16 274
23 Nov-16 237
300
24 Dec-16 278
25 Jan-17 284 250
26 Feb-17 277
27 Mar-17 317 200
28 Apr-17 313
29 May-17 318 150
30 Jun-17 374
2017 16 16 16 16 16 16 16 16 16 16 16 16 17 17 17 17 17 17 17 17 17 17 17 17
31 Jul-17 413 n- b- r- r- y- n- l- g- p- t- v- c- n- b- r- r- y- n- l- g- p- t- v- c-
32 Aug-17 405 Ja Fe Ma Ap Ma Ju Ju Au Se Oc No De Ja Fe Ma Ap Ma Ju Ju Au Se Oc No De
33 Sep-17 355
34 Oct-17 306
35 Nov-17 271
36 Dec-17 306
37 Jan-18
38 Feb-18
39 Mar-18
40 Apr-18
41 May-18
42 Jun-18
2018 ?
43 Jul-18
44 Aug-18
45 Sep-18
46 Oct-18
47 Nov-18
48 Dec-18
Fifth data: Running Status
Running time
Date KM
(in minutes)
15-Oct-23 70 10
28-Nov-23 70 10
16-Oct-23 35 5
17-Oct-23 72 10 30-Nov-23 35 5
18-Oct-23 73 10
19-Oct-23 67 10 2-Dec-23 72 10
20-Oct-23 65 10
21-Oct-23 90 15 4-Dec-23
22-Oct-23 62 10
5-Dec-23
23-Oct-23 64 10
24-Oct-23 65 10 6-Dec-23
25-Oct-23 66 10
26-Oct-23 63 10 7-Dec-23
27-Oct-23 62 10
28-Oct-23 8-Dec-23
29-Oct-23
30-Oct-23 65 10
10-Dec-23 42.195 km
31-Oct-23 63 10 12-Dec-23
1-Nov-23 62 10
2-Nov-23 63 10 14-Dec-23
3-Nov-23 62 10
4-Nov-23 63 10 16-Dec-23
5-Nov-23 65 10
17-Dec-23 173 25
6-Nov-23 65 10
7-Nov-23 64 10
18-Dec-23
8-Nov-23 65 10
9-Nov-23 130 20 20-Dec-23 32 5
10-Nov-23
11-Nov-23 62 10 22-Dec-23 60 10
12-Nov-23
13-Nov-23 62 10 24-Dec-23 35 5
14-Nov-23 63
26-Dec-23 65 10
15-Nov-23 130 22
16-Nov-23 62 10 28-Dec-23
17-Nov-23
18-Nov-23 30-Dec-23
19-Nov-23 67 10
20-Nov-23 68 10 31-Dec-23 60 10
21-Nov-23
1-Jan-24 32 5
22-Nov-23 62 10
23-Nov-23 30 5 2-Jan-24 60 10
24-Nov-23 126 18.48
24-Nov-23
26-Nov-23 62 10
Running Graph

200

180

160

140

120

100

80

60

40

20

0
11-Oct-23 21-Oct-23 31-Oct-23 10-Nov-23 20-Nov-23 30-Nov-23 10-Dec-23 20-Dec-23 30-Dec-23 9-Jan-24

Running time (in minutes) KM

 Quantitative Methods

Actual Demand Passenger Vehicle Sales Air passenger movements (in

(×100000) 5000000 Lakhs)
30 4500000 450
28 4000000
400
26 3500000
24 350
3000000
22 300
20 2500000
250
18 2000000
16 1500000 200
14 150
12 1000000
10 500000 16 16 16 16 16 16 17 17 17 17 17 17
0 an- ar- ay- Jul- ep- ov- an- ar- ay- Jul- ep- ov-
14 15 16 17 18 19 20 21 22 23 J M M S N J M M S N
20 20 20 20 20 20 20 20 20 20 2000 2005 2010 2015 2020 2025

Level Level +Trend Level + Trend+ Seasonality

Do you think forecasting approach will be same for all type of products
Quantitative Methods of forecasting

Demand Forecasting
(Independent demand)

Time series Causal Analysis/

Associative
model

Naive Averaging Trend Seasonality

 Average
 Simple moving average  Regression analysis  Regression
 Weightage average  Holt’s exponential
 Exponential smoothing smoothing
Method 1 : The Naïve
Method
● Ft=A t-1 : Simplest Approach to Forecasting

Forecast of Demand Period Demand Forecast

in Period 2
(This forecast made 1 130 -
After seeing demand
in period 1) 2 155 130
3 145 155
Forecast of Demand
in Period 5
4 160 145
(This forecast made 5 151 160
After seeing demand
in period 4) 6 143 151

7 143
Method 2 : The Simple
Average ●
Ft=(A1+ A2+A3+…At-1)/(t-1)

Period Demand Forecast

1 130 -
2 155 130/1 =130

3 145 (130+155)/2 =142.50

4 160 (130+155+145)/3 = 143.33

5 151 (130+155+145+160)/4 = 147.5

6 143 (130+155+145+160+151)/5 = 148.2

7 (130+155+145+160+151+143)/6 = 147.33
The Moving Average
Forecast
“The recent history is more relevant ”
● The Simple Average forecast uses ALL THE HISTORY of demands to
generate the forecast for the next period

● The (Simple) Moving Average forecast (order n) uses ONLY THE n

MOST RECENT period demands to generate the forecast for the next
period
Method 3 : The Simple Moving
Average
● Forecast for period t = the average of demand in the
past n periods (from period t-1 to t-n)

A t- 1+ A t- 2 +A t- 3+...+A t-n
F t=
n

● At-1 is the actual sales in period t-1

Method 3 : The Simple Moving Average(Cont’d)

Period Demand Forecast Using Using

n=3 n=4
1 130 -

2 155 - Not enough history

3 145 - Not enough history

4 160 143.33 (130+155+145)/3 = 143.33

5 151 153.33 (155+145+160)/3 = 153.33

6 143 152.00 (145+160+151)/3 =

152.00

7 151.33 (160+151+143) /3 = 151.33

Method 3 : The Simple Moving Average
(Cont’d)

165

160

155

150

145

140

135

130

125

120
1 2 3 4 5 6 7

D eman 2 3 4
d period period period

● The four-week average is smoother than three-week

average.
 The Weighted Moving Average Forecast
 The (simple) Moving Average forecast (order n) treats each
of the n most recent demands EQUALLY in generating the
forecast for the next period

 The Weighted Moving Average forecast (order n)

weights each of the n most recent demands (possibly)
DIFFERENTLY in generating the forecast for the next
period
Method 4: Weighted Moving Average
(WMA)
● When a detectable trend or pattern is present, weights can
be used to place more emphasis on recent values

● Weights based onintuition

 Weights are values between 0 and 1
 Weights sum to 1.0
 Weights impact stability and responsiveness of the forecast

● Formula

Ft= w1At-1
t-1+ w22 At-2 +w33A
t-2+w At-3t-3+...+w
+...+wnnAt-n
At-n
Method 4 : Weighted Moving Average
(WMA)
n=3:
Period Demand Forecast w t-1 =0.5, t-2=0.3,
w
1 130 - wt-3=0.2

2 155 - Not enough history

3 145 - Not enough history

4 160 145.00 0.2(130)+0.3(155)+0.5(145) =145.00

5 151 154.50 0.2(155)+0.3(145)+0.5(160) =154.50

6 143 152.20 0.2(145)+0.3(160)+0.5(151) =152.50

7 148.80 0.2(160)+0.3(151)+0.5(143) =148.80

Forecast Errors
Forecast error is the difference between the forecast
value and what actually occurred.
 Forecast errort = Actualt –Forecastt
 et =At − Ft

Measures of Error
 Mean absolute deviation (MAD)
 Mean absolute percent error (MAPE)
 Mean squared error (MSE)
 Root mean sum of square (RMSE)
Measuring Forecast Errors: Mean Absolute
Deviation (MAD)
n
n
 A -F
t t

MAD =
MAD = 
t=1 A - F
n
t t
t=1
n
 The ideal Mean Absolute Deviation (MAD)is zero which would mean
there is no forecasting error at all.

 The larger the MAD, the less the accurate the resulting model.
Measuring Forecast Errors: Mean
Absolute Deviation (MAD)

Absolut
Period Demand Forecast Error e Error
1 130 - - -

2 155 130.00 25.00 25.00

3 145 155.00 -10.00 10.00

4 160 145.00 15.00 15.00

5 151 160.00 -9.00 9.00

6 143 151.00 -8.00 8.00

MAD = 13.40
Measuring Forecast Errors: Mean Squared Error
(MSE)

Squared
Period Demand Forecast Error Error

1 130 - - -
2 155 130.00 25.00 625.00
3 145 155.00 -10.00 100.00
4 160 145.00 15.00 225.00
5 151 160.00 -9.00 81.00
6 143 151.00 -8.00 64.00
MSE = 219.00
Measuring Forecast Errors: Mean Absolute
Percentage Error (MAPE)

Abs
Period Demand Forecast Error %
%
Error
Error
1 130 - - - -
2 155 130.00 25.00 =25/155=16.13% 16.13%
3 145 155.00 -10.00 =-10/145=-6.90% 6.9%
4 160 145.00 15.00 =15/160=9.38% 9.38%
5 151 160.00 -9.00 =-9/151=-5.96% 5.96%
6 143 151.00 -8.00 =-8/143=-5.59% 5.59%
MAPE 8.79%
Measure of Forecast Error
Mean Absolute deviation (MSD): 𝑀A𝐷 =

Mean squared error (MSE): 𝑀A𝐷 =

The MSE penalizes large errors more significantly than small efforts because
all errors are squared.
Lower MSE implies better prediction

Root Mean squared error (RMSE): =

Mean Absolute percentage error (MAPE): ×100

MAPE is dimensionless it can be used for comparing different models with

varying scales.
It is a good measure of forecast error when forecast error when the underlying
forecast has significant seasonality and demand varies considerably from one
period to the next
MSE and RMSE are the popular forecasting accuracy indicators
Problem 1
Period Demand
1 20 a) If a three-period moving average had been used to forecast sales,
2 24
3 24 what would the daily forecasts have been starting with the
4 16
5 20 forecast for Day 4?
6
7
20
17 b) If a five-period moving average had been used
8
9
22
22 determine what the forecasts would have been for each day,
10
11
20
21
starting with Day 6.
12
13
16
18
c) Use a three-period weighted moving average with w1=0.2,
14
15
17
16
w2=0.3, and w3=0.5 and forecast sales for the 4th day onwards.
16
17
19
23
d) Use a four-period weighted moving average with w1=0.1,
18
19
20
23
w2=0.2, w3=0.3, and w4=0.4 and forecast sales for the 5th day
20
21
17
22
onwards.
22
23
18
15
e) Plot the original data and each set of forecasts on the same
24
25
22
23
graph. Which forecast approach is better?
26 18
27 16
28 22
29 20
30 24
31 ?
 Which approach is better ?
 Sensitivity to Trends: WMAs are typically more sensitive to recent trends since they
assign more weight to the latest data points.

 Data Volatility: SMAs can smooth out short-term fluctuations and highlight longer-term
trends in data that is very volatile. This can be beneficial when you want to avoid reacting
to what might be considered "noise" in the data.

 Data Patterns: If the data has a seasonal pattern or other cyclical changes, neither SMA
nor WMA may be sufficient as they don't inherently account for such patterns. More
sophisticated methods like Exponential Smoothing or ARIMA may be more suitable in
such cases.
Practice Problem-1
Day Number Sold
a) If a two-period moving average had been used to
1 25 forecast sales, what would the daily forecasts have
2 31 been starting with the forecast for Day 3?
3 29 b) If a four-period moving average had been used
4 33 determine what the forecasts would have been for
5 34 each day, starting with Day 5.
6 37 c) Use a three-period weighted moving average with
7 35 w1=0.2, w2=0.3, and w3=0.5 and forecast sales for
8 32 the 4th day onwards.
9 38 d) Use a four-period weighted moving average with
10 40 w1=0.4, w2=0.3, w3=0.2, and w4=0.1and forecast
11 37 sales for the 5th day onwards.
12 32 e) Plot the original data and each set of forecasts on
the same graph. Which forecast has the better
ability to respond quickly to changes?
Method 5 : Exponential
Smoothing
Smoothing constant

● Ft = Ft-1 + α(At-1 -Ft-1)= α(At-1 )+(1- α)Ft-1

Forecast= Previous forecast + α(previous actual sales- previous forecast )

 Ft is the forecast for the period t
 At-1 is the actual sales in the previous period
 Ft-1 is the forecast for the previous period
 α: ranges from 0 to 1 and is subjectively chosen
Problem 2
Period Demand
1 20
Day Number Sold 2 24
1 25 3 24
4 16
2 31 5 20
6 20
3 29
7 17
4 33 8 22
9 22
5 34 10 20
6 37 11 21
12 16
7 35 13 18
8 32 14 17
15 16
9 38 16 19
17 23
10 40
18 20
11 37 19 23
20 17
12 32
21 22
13 22 18
23 15
24 22
25 23
26 18
27 16
28 22
29 20
30 24
31 ?
Method 5 : Exponential Smoothing
(Cont’d)
● Forecast effects of Smoothing Constant α

Weights on Actual Sales

Prior Period 2 PeriodsAgo 3 PeriodsAgo

α α(1 − α) α(1 − α)2

α = 0.10 0.271
α = 0.90 0.999
Summary: Exponential Smoothing
Most popular
● Because…
 Formulating an exponential model is relatively easy
and it is surprisingly accurate.
 Little computation is required so the computer
storage requirements are small

● It is a special form of weighted moving average

 Weights decline exponentially with most recent
data weighted most
Regression
Passenger
Period Year Vehicle
Sales
1 2004 1061572
2 2005 1143076 Passenger Vehicle Sales
3 2006 1379979 5000000
4 2007 1549882 4500000
5 2008 1552703
4000000
6 2009 1951333
3500000
7 2010 2501542
8 2011 2629839 2531999 3000000

2012 2665015 2500000

9
10 2013 2503509 2000000

11 2014 2601111 1500000

12 2015 2789678 1000000
13 2016 3047582 500000
14 2017 3288581 0
15 2018 3377389 2000 2005 2010 2015 2020 2025

16 2019 2773575
17 2020 3062280
18 2021 3650698
19 2022 4578639
20 2023 ?
Trend-Adjusted Exponential Smoothing (Holt’s model)
The trend adjusted forecast consists of two components
 Smoothed error
 Trend factor

𝐹𝑡+1 = 𝐿𝑡 + 𝑇𝑡
𝐿𝑡 = 𝛼𝐷𝑡 + 1 − 𝛼 𝐿𝑡−1 + 𝑇𝑡−1
𝑇𝑡 = 𝛽 𝐿𝑡 − 𝐿𝑡−1 + 1 − 𝛽 𝑇𝑡−1

𝛼 and 𝛽 are smoothing constants

0≤α, 𝛽 ≤1
Trend-adjusted exponential smoothing
has the ability to respond to changes in
trend
 Trend-Adjusted Exponential Smoothing (Holt’s model)
Period t Dt Lt Tt Ft Regression line
1 26 24.27 1.69
2 28

3 29

4 31

5 32

6 35

𝐹𝑡+1 = 𝐿𝑡 + 𝑇𝑡
𝐿𝑡 = 𝛼𝐷𝑡 + 1 − 𝛼 𝐿𝑡−1 + 𝑇𝑡−1
𝑇𝑡 = 𝛽 𝐿𝑡 − 𝐿𝑡−1 + 1 − 𝛽 𝑇𝑡−1
Trend-Adjusted Exponential Smoothing (Holt’s model)

Period t Dt Lt Tt Ft

1 26 24.27 1.69
2 28 26.36 1.81 25.95

3 29 28.34 1.86 28.17

4 31 30.36 1.91 30.19

5 32 32.21 1.89 32.26

6 35 34.28 1.94 34.10

7 36.23
Problem
Passenger Lt Tt Ft
Period Year Vehicle
Sales
1 2004 1061572
2 2005 1143076
3 2006 1379979
4 2007 1549882
5 2008 1552703
6 2009 1951333
7 2010 2501542
8 2011 2629839
9 2012 2665015
10 2013 2503509
11 2014 2601111
12 2015 2789678
13 2016 3047582
14 2017 3288581
15 2018 3377389
16 2019 2773575
17 2020 3062280
18 2021 3650698
19 2022 4578639
Problem

Air transport, passengers carried data – India (1970-2020)

Develop the forecast model using Holt’s and regression-
based approach.
 Find the optimal values of 𝛼 and 𝛽 using solver.
 Compare these two methods using MSE and suggest
which method is more appropriate.
 What could be an alternate approach?
 Time series data with seasonality
 Forecasting data with seasonality is an essential aspect in various fields, including sales,
marketing, and even academic research

 Seasonality refers to periodic fluctuations that regularly occur in data due to seasonal
factors. It's often seen in monthly or quarterly sales data, where certain times of the year
are consistently higher or lower than others.
 Key Steps in Seasonal Forecasting
1. Identifying Seasonality: This involves analyzing your data to determine if there is a seasonal pattern. This can
be done through visual examination of time series plots or using statistical tests for seasonality.

2. Decomposition of Time Series: Time series data can be decomposed into three components: trend, seasonality,
and randomness. This helps in understanding the underlying patterns. Methods like STL (Seasonal and Trend
decomposition using Loess) are commonly used.

3. Choosing the Right Model: Depending on the nature of the seasonality, different models can be applied.
Common models include:
1. Seasonal Factor
2. Regression Approach
3. SARIMA (Seasonal ARIMA): An extension of ARIMA that specifically addresses seasonality.
4. Exponential Smoothing: Methods like Holt-Winters which are simple yet powerful for forecasting
seasonal data.

4. Parameter Selection: This involves choosing parameters that best fit the model to your data. For SARIMA,
these include seasonal order and non-seasonal order parameters.

5. Model Validation: Before using the model for forecasting, it’s crucial to validate it using historical data. This
involves checking the model’s performance against known data and adjusting as necessary.

6. Forecasting: Once the model is validated, it can be used to forecast future data points.
Time series data with seasonality
Sales Year 1 Year 2 Year 3 Year 4 Year 5 25,000
Demand Dt
JAN 2,000 3,000 2,000 5,000 5,000 20,000

FEB 3,000 4,000 5,000 4,000 2,000 15,000

MAR 3,000 3,000 5,000 4,000 3,000 10,000

APR 3,000 5,000 3,000 2,000 2,000

5,000

MAY 4,000 5,000 4,000 5,000 7,000

-
0 10 20 30 40 50 60 70
JUN 6,000 8,000 6,000 7,000 6,000

JUL 7,000 3,000 7,000 10,000 8,000

AUG 6,000 8,000 10,000 14,000 10,000

SEP 10,000 12,000 15,000 16,000 20,000

OCT 12,000 12,000 15,000 16,000 20,000

NOV 14,000 16,000 18,000 20,000 22,000

DEC 8,000 10,000 8,000 12,000 8,000

Regression Approach with help of dummy coding

Sales Year 1 Year 2 Year 3 Year 4 Year 5

JAN 2,000 3,000 2,000 5,000 5,000

FEB 3,000 4,000 5,000 4,000 2,000

MAR 3,000 3,000 5,000 4,000 3,000

APR 3,000 5,000 3,000 2,000 2,000

MAY 4,000 5,000 4,000 5,000 7,000

JUN 6,000 8,000 6,000 7,000 6,000

JUL 7,000 3,000 7,000 10,000 8,000

AUG 6,000 8,000 10,000 14,000 10,000

SEP 10,000 12,000 15,000 16,000 20,000

OCT 12,000 12,000 15,000 16,000 20,000

NOV 14,000 16,000 18,000 20,000 22,000

DEC 8,000 10,000 8,000 12,000 8,000

 Solve this problem with help of dummy coding and estimate month wise sales
for Year 6
Python
Causal Analysis/ Associative model

 Multiple regression
Store Price ($) Advertising ($ '000s) Burger Sales (units '000s)

Baulkham Hills 7.5 3.0 360

Bella Vista 4.5 3.0 520
Blacktown 8.0 2.7 240
Castle Hill 6.8 3.0 350

Claremont Meadows 7.5 3.3 350

Dee Why 5.1 4.2 820

Forestville 5.0 3.5 500
Frenchs Forest 7.0 2.7 300
Glenwood 8.0 2.0 230
Killara 7.2 3.5 310
Macquarie Park 6.4 3.7 430
North Ryde 7.0 3.5 400
Penrith 5.5 3.3 350
Rhodes 5.0 4.0 630
West Pennant Hills 7.9 2.7 250
Wynyard 5.9 4.0 440
Case Article
Forecasting
Beer Demand
at Anadolu
Efes
This case won first prize in the 2007
Case Competition of the Institute
for Operations Research and the
Management Sciences (INFORMS)
 Synopsis of the Case Study
 Efes Beverage Group is the beverage division of one of Turkey’s leading corporations
and it has a share of about 78% in the Turkish beer market.

 Efes forecasts the monthly demand for the coming year during the fall of the current
year.

 Historically, high-level sales managers, based on input from their sales personnel, have
done this forecasting mostly subjectively.

 They now want to formalize this process so that significant factors on beer demand can
be identified and used to predict the monthly demand
 Questions
 Plot the monthly beer demand and discuss your observations.
 Run the multiple linear regression model to explain the monthly beer demand in terms
of the predictor variables, discuss the validity of the model, and interpret the results.
 Are there any problems with the validity of the original model? If so, make the
necessary modifications and repeat the process with new model(s). Employ variable
selection methods to eliminate irrelevant variables.
 Are there unusual observations? Interpret and discuss.
 Discuss the predictive capability of each predictor variable. Make predictions for the
monthly demands of the following year. Make necessary assumptions if you need data
on predictor variables. You may create scenarios by assuming some values for
“uncontrollable” variables and trying some values for “controllable” variables. Discuss
the results.
 Are there alternative models that include different sets of predictor variables with
approximately the same explanation power? How would you interpret such results?
Quantitative Methods of forecasting
Demand Forecasting
(Independent demand)

Time series Causal Analysis/

Associative model

Naive Averaging Trend Seasonality

 Average  Regression
 Regression analysis
 Simple moving average
 Holt’s exponential
 Weightage average
smoothing
 Exponential smoothing