Session 4 - 11 - Demand Forecasting

MBA – Batch 2021-23
Semester - I
MB700.4 MANAGERIAL ECONOMICS
Session 3 - 10
•Demand Estimation / Forecasting
• Meaning
• Importance
• Features
• Application
• Objectives
•Qualitative Methods of Forecasting
• Definition
• Conditions for Use
• Types of Qualitative Methods
•Quantitative Methods of Forecasting
• Definition
• Conditions for Use
• Types of Quantitative Methods
Demand Estimation / Forecasting
• In Demand forecasting managers forecast the most likely future
demand of a product so that he can make necessary arrangement
for the various factor of production i.e labor, raw material,
machines, money etc.
• Demand forecasting tells the expected level of demand at some
future date on the basis of past and present information.
• It helped in production planning, new product development,
capacity enhancement or new schemes etc. Demand forecasting is
generally used for short term estimation (6 months to one and half
year) as well as long term forecasting (2 – 5 years).
• The process of demand estimation/forecasting can be broken into
two parts i.e. analysis of the past conditions and analysis of current
conditions with reference to a probable future trend. It helps in
estimating the most likely demand of a good or service under given
business conditions means various factors affecting business.
2
Managerial Economics - Mr. Snehal Bhatt
Importance of Demand Forecasting
• As business is done in an uncertain and risky business
environment and managers have to take decisions under
uncertain and risky conditions.
• Demand forecasting help the managers in forecasting the most
likely future sale of their products, accordingly manager plan their
production, arrange various inputs like labour, material, capital
and techniques etc. prepare future budgets (Expenditure and
Revenue) and formulate various marketing and supply chain
strategies or policies to achieve the budgeted targets.
• This will help up to certain extent in managing the future risks
caused due to varied / changing business conditions as well as
in optimum utilization of available business resources.
3
Features of Demand Forecasting
• Demand Forecasting is a process to investigate and measure the
forces / factors that determine sales for existing and new
products.
• It is an estimation of most likely future demand for a product
under given business conditions.
• It is basically an educated and well thought out guesswork in
terms of specific quantities.
• Demand Forecasting is done for a specific period of time (i.e. the
sufficient time required to take a decision and put it into action).
• It tells us only the approximate expected future demand for a
product based on certain assumptions and cannot be 100%
precise.
4
Applications of Demand Forecasting
• Demand Forecasting is important in the field of marketing as it
will help to launch correct product at correct time.
• It is an also helpful in the field of accounting. Good forecast helps
in appropriate production planning, process selection, capacity
planning, facility layout planning, and inventory management, etc.
• Forecasting human resource demand is the process of estimating
the future human resource requirement of right quality and right
number.
5
Objectives of Demand Forecasting
6
What is forecasting all about?
Demand for Mercedes E Class We try to predict the

future by looking back
at the past
Predicted
demand
looking
Time back six
Jan Feb Mar Apr May Jun Jul Aug months
Actual demand (past sales)

Forecasted / Predicted demand
7
Types of Forecasting Methods
Qualitative Methods Quantitative Methods
Rely on subjective Rely on data and

opinions from experts. analytical techniques.
https://www.youtube.com/watch?v=fp-1_9mLlbc https://www.youtube.com/watch?v=-fgGGaBGt24
8
Qualitative Methods of Demand Forecasting
• The qualitative method is when you forecast demand when there
is no prior data or sales numbers to work with by using the
opinions of a group of experts. The data that you collect from
your group of experts is then used to predict a possible quantity
of future demand.
• Qualitative Methods offer advantages and disadvantages. When
statistics are unavailable, Qualitative Analysis is to be used.
Qualitative analysis allows more flexibility and creativity, making
it a useful tool for small business. It can be used to identify
problems, such as why a product is not selling well, or even as a
forecasting technique.
• Qualitative analysis relies on personal viewpoints and

experiences, including educated guesses, focus groups, panels
and questionnaires.
9
Qualitative Methods of Demand Forecasting
10
Procedure of Delphi Method:
Round Expert 1 Expert 2 Expert 3 Expert 4 Average

1 100 150 125 80 114
2 105 120 120 100 112
3 108 115 115 108 112
4 110 112 112 110 111
5 111 111 111 111 111
11
Quantitative Methods of Demand Forecasting
• Quantitative Techniques use the data gathered over time and use
statistical techniques to come up with a forecast.
Quantitative methods are more objective and ‘scientific’ . It
always involve the historical data, and by using the mathematical
models to process those information to found out the patterns
embed in the data.
• Exposes Patterns Numerical data can clearly expose patterns of
spending, sales, and scheduling within the business. This type
of forecasting clearly shows trends over a specific time period
and whether these patterns are consistent from year to year.
12
Moving Average Method
• The moving average method uses the last t periods in order to

predict demand in period t+1.
• There can be two types of moving average methods: Simple
Moving Average and Weighted Moving Average
• The moving average model assumption is that the most
accurate prediction of future demand is a simple (linear)
combination of past demand.
t-n t-2 t-1 t t+1
13
Method – 1: Simple Moving Average Method
In the simple moving average method the forecast value is
At + At-1 + … + At-n
Ft+1 =
n
t is the current period.

Ft+1 is the forecast for next period
n is the forecasting horizon (how far back we look),
A is the actual sales figure from each period.
https://www.youtube.com/watch?v=Wo5YWXDRXv8
14
Example-1:
Month Bottles
Jan (t – 5) 1,325
Feb (t – 4) 1,353
Mar (t – 3) 1,305
Apr (t – 2) 1,275
May (t – 1) 1,210
Jun (t) 1,195 What will
Jul (t + 1) ? the sales be
for July?
15
What if we use a 3-month simple moving average?
AJun + AMay + AApr

FJul = = 1,227
3
What if we use a 5-month simple moving average?
AJun + AMay + AApr + AMar + AFeb

FJul = = 1,268
5
16
What do we observe?
5-month average smoothes data more;

3-month average more responsive
17
Method – 1: Simple Moving Average Method
In the simple moving average method the forecast value is
At + At-1 + … + At-n
Ft+1 =
n

https://www.youtube.com/watch?v=Wo5YWXDRXv8
18
Example-2:
Actual
2 Month Error = 3 Month Error = 4 Month Error = 5 Month Error =
Month Sales
MA (F) A - F MA (F) A-F MA (F) A-F MA (F) A–F
(A)
January 1,220
February 1,295
March 1,350
April 1,450
May 1,525
June 1,600
July 1,675
August 1,750
September 1,800
October 1,925
November ?
19
Solution-2:
20
Revision – Practice Question 1 – Moving Average with Error
Actual
3-MA Error = 5-MA Error = 7-MA Error = 10-MA Error = 12-MA Error =
Year Sales
(F) A-F (F) A-F (F) A-F (F) A-F (F) A-F
(A)
2000 2455
2001 2480
2002 2419
2003 2569
2004 2487
2005 2670
2006 2676
2007 2863
2008 2945
2009 3101
2010 3209
2011 3458
2012 3176
2013 3281
2014 3322
2015 3277
2016 3531
2017 3628
2018 3738
2019 3755
2020 ?
Total Error 21
Revision – Solution Question 1 – Moving Average with Error
Actual
3-MA Error = 5-MA Error = 7-MA Error = 10-MA Error = 12-MA Error =
Year Sales
(F) A-F (F) A-F (F) A-F (F) A-F (F) A-F
(A)
2000 2455
2001 2480
2002 2419
2003 2569 2452 117
2004 2487 2490 -3
2005 2670 2492 178 2482 188
2006 2676 2576 100 2525 151
2007 2863 2611 252 2565 298 2537 326
2008 2945 2737 208 2653 292 2595 350
2009 3101 2828 273 2729 372 2662 439
2010 3209 2970 239 2851 358 2759 450 2667 542
2011 3458 3085 373 2959 499 2851 607 2742 716
2012 3176 3256 -80 3116 60 2989 187 2840 336 2778 398
2013 3281 3281 0 3178 103 3062 219 2916 365 2838 443
2014 3322 3305 17 3245 77 3148 174 2987 335 2905 417
2015 3277 3260 17 3290 -13 3214 63 3071 206 2980 297
2016 3531 3294 237 3303 228 3261 270 3131 400 3039 492
2017 3628 3377 251 3318 310 3322 306 3217 411 3126 502
2018 3738 3479 259 3408 330 3382 356 3293 445 3206 532
2019 3755 3633 122 3500 255 3422 333 3373 382 3295 460
2020 ? 3707 3586 3505 3438 3369
Total Error 2560 3508 4080 4138 3541
Mean Error 150.59 233.87 313.85 413.80 442.63
22
Look the data again.. Is there any trend?
Month Bottles
Jan 1,325
Feb 1,353
Mar 1,305
Apr 1,275
May 1,210 What will
Jun 1,195 the sales be
Jul ? for July if
we use 6
Month MA?
23
Why do we need the WMA models?
Because of the ability to give more importance to what

happened recently, without losing the impact of the past.
Demand for Mercedes E-class Actual demand (past sales)

Prediction when using 6-month SMA
Prediction when using 6-months WMA
1277 For a 6-month

SMA, attributing
equal weights to all
past data we miss
Time the downward trend
Jan Feb Mar Apr May Jun Jul Aug
24
Method – 2: Weighted Moving Average Method
We may want to give more importance to some of the data…
Ft+1 = wt At + wt-1 At-1 + … + wt-n At-n
wt + wt-1 + … + wt-n = 1
w is the importance (weight) we give to each period
https://www.youtube.com/watch?v=DipOB2H6ick
25
Example-3
Month Bottles
Jan 1,325
Feb 1,353 What will
Mar 1,305 be the
Apr 1,275
sales for
July?
May 1,210
Jun 1,195
Jul ?
26
6-Month Moving Average V/ Weighted Moving Average
In other words, because we used equal weights in Moving

Average, a slight downward trend that actually exists is not
observed. Hence, less weightage should be given to old data points
and maximum weightage should be given to most recent data
point. 27
6-Month Moving Average V/ Weighted Moving Average
In other words, because we used equal weights in Moving

Average, a slight downward trend that actually exists is not
observed. Hence, less weightage should be given to old data points
and maximum weightage should be given to most recent data
point. 28
What if we use a Weighted Moving Average?
Make the weights for the last five months less than the most
recent month.
6-month WMA WMA WMA

SMA 10% / 50% 25% / 75% 15% / 85%
July
1,277 1,244 1,220 1,177
Forecast
The higher the importance we give to recent data, the more we

pick up the declining trend in our forecast.
29
Method – 3: Exponential Smoothing Method
Main idea: The prediction of the future depends mostly on the

most recent observation, and on the error for the latest forecast.
Smoothing
constant
Denotes the importance
alpha α of the past error
30
Why use Exponential Smoothing Method?
1. Uses less storage space for data

2. Extremely accurate
3. Easy to understand
4. Little calculation complexity
5. There are simple accuracy tests
https://www.youtube.com/watch?v=k_HN0wOKDd0
31
Exponential Smoothing Method
Assume that we are currently in period (t+1). We calculated the
forecast for the last period (Ft) and we know the actual demand
last period (Dt) …
Ft+1 (January) = αDt (Actual Demand of December)+ (1-α)Ft (Forecast of December)
The smoothing constant α expresses how much our forecast will

react to observed difference. Value of alpha used by
Pharmaceutical
Value of α varies from 0 to 1. Value of alpha used by
Automobile
If α is low: there is little reaction to differences.
If α is high: there is a lot of reaction to differences.
In Moving Averages the past observations are weighted equally, Exponential
Smoothing assigns exponentially decreasing weights as the observation get
older. In other words, recent observations are given relatively more weight in
32
forecasting than the older observations. Managerial Economics - Mr. Snehal Bhatt
Example-4:
 = 0.2
Month Actual
Jan 1,325
Feb 1,353
Mar 1,305
Apr 1,275
May 1,210
Jun ?
33
Solution-4:
 = 0.2
Ft+1 = αDt+ (1-α)Ft

34
Example-5:
 = 0.8
Month Actual
Jan 1,325
Feb 1,353
Mar 1,305
Apr 1,275
May 1,210
Jun ?
35
Example-5a:  = 0.55
Month Actual
Jan 1,300
Feb 1,750
Mar 1,500
Apr 1,650
May 1,900
Jun 1,870
July ?
36
Solution-5:
 = 0.8
37
Example-6: Month
Actual
Sales
 = 0.8
Error
 = 0.5
Error
 = 0.3
Error
(A)
January 1,220
February 1,295
March 1,350
April 1,450
May 1,525
June 1,600
July 1,675
August 1,750
September 1,800
October 1,925
November ?
Total Absolute Error
MAD 38
Solution-6:
Month Actual a
= a= a=
0.8 Error 0.5 Error 0.3 Error
January 1,220 1220 NA 1220 NA 1220 NA
February 1,295 1220 75 1220 75 1220 75
March 1,350 1280 70 1258 92.5 1243 108
April 1,450 1336 114 1304 146 1275 175
May 1,525 1427 98 1377 148 1327 198
June 1,600 1505 95 1451 149 1387 213
July 1,675 1581 94 1525 150 1451 224
August 1,750 1656 94 1600 150 1518 232
September 1,800 1731 69 1675 125 1588 212
October 1,925 1786 139 1738 187 1651 274
November ? 1897 NA 1831 NA 1733 NA
Total Absolute
847 1223 1711
Error
MAD 94.06 135.84 190.15
39
Method – 4: Naive Method
Main Idea: Estimating technique in which the last period's

actuals are used as this period's forecast, without adjusting them
or attempting to establish causal factors.
Only
important
data point is
most recent
past
demand
40
Example-7
Month Actual Forecast Error

Jan 1,325 - NA
Feb 1,353 1,325 1353-1325 = 28
Mar 1,305 1,353 1305-1350 = -48
Apr 1,275 1,305 1275-1305 = -30
May 1,210 1,275 1210-1275 = -65
Jun ? 1,210 NA
Mean Error (ME) -115/4 = -28.75
Mean Absolute Deviation (MAD) 171/4 =42.75
41
How can we compare across Forecasting Methods?
We need a metric that provides estimation of accuracy
Errors can be:

Forecast Error 1. Biased (consistent)
2. Random
Forecast error = Difference between actual and forecasted value

(also known as residual)
42
Measuring Accuracy: ME
ME = Mean Error (Bias)

It is the average error in the observations
A F t t
MFE  i 1
n
1. A more positive or negative MFE implies worse

performance; the forecast is biased.
43
Measuring Accuracy: MAD
MAD = Mean Absolute Deviation

It is the average absolute error in the observations
 A F t t
MAD  i1
n
1. Higher MAD implies worse performance.

2. If errors are normally distributed, then σε=1.25MAD
44
Key Point
Forecast must be measured for accuracy!
The most common means of doing so is by

measuring the either the Mean Absolute
Deviation (MAD)or the standard deviation of
the forecast error
45
Example – 8:
Exponential Smoothing ( = 0.2)
Error Absolute
Month Actual At Forecast Ft
= At - Ft Error
Jan 1,325 1,325 NA NA
Feb 1,353 1,325 28 28
Mar 1,305 1,331 -26 26
Apr 1,275 1,325 -50 50
May 1,210 1,315 -105 105
Jun ? 1,294 NA NA
Mean Error (ME) / Mean

-38.25 52.75
Absolute Deviation (MAD)
46

Session 4 - 11 - Demand Forecasting

Uploaded by

Copyright:

Available Formats

Session 4 - 11 - Demand Forecasting

Uploaded by

Document Information

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

Session 4 - 11 - Demand Forecasting

Uploaded by

Copyright:

Available Formats

MBA – Batch 2021-23

Demand for Mercedes E Class We try to predict the

Actual demand (past sales)

Rely on subjective Rely on data and

• Qualitative analysis relies on personal viewpoints and

Round Expert 1 Expert 2 Expert 3 Expert 4 Average

• The moving average method uses the last t periods in order to

t-n t-2 t-1 t t+1

In the simple moving average method the forecast value is

t is the current period.

AJun + AMay + AApr

What if we use a 5-month simple moving average?

AJun + AMay + AApr + AMar + AFeb

5-month average smoothes data more;

In the simple moving average method the forecast value is

t is the current period.

Because of the ability to give more importance to what

Demand for Mercedes E-class Actual demand (past sales)

1277 For a 6-month

Ft+1 = wt At + wt-1 At-1 + … + wt-n At-n

In other words, because we used equal weights in Moving

In other words, because we used equal weights in Moving

6-month WMA WMA WMA

The higher the importance we give to recent data, the more we

Main idea: The prediction of the future depends mostly on the

1. Uses less storage space for data

The smoothing constant α expresses how much our forecast will

Ft+1 = αDt+ (1-α)Ft

Total Absolute Error

Main Idea: Estimating technique in which the last period's

Month Actual Forecast Error

We need a metric that provides estimation of accuracy

Errors can be:

Forecast error = Difference between actual and forecasted value

ME = Mean Error (Bias)

1. A more positive or negative MFE implies worse

MAD = Mean Absolute Deviation

1. Higher MAD implies worse performance.

Forecast must be measured for accuracy!

The most common means of doing so is by

Feb 1,353 1,325 28 28

Mar 1,305 1,331 -26 26

Apr 1,275 1,325 -50 50

May 1,210 1,315 -105 105

Mean Error (ME) / Mean

You might also like