Logistic

Management
 Forecasting – A Case Study

Project Background

A client ships products across the globe from its operations point in
Asia and caters to a number of product categories. The product
categories are well diversified and has different demand dynamics
including global trends and seasonality, destination country, macro-
economic factors etc.

Analytics Objectives

To create a comprehensive system for demand forecasting across all

products/geographies of future forecast performance.
To provide statistical measures of the forecast performance
 Forecasting: A Case Study

Solution

• Developed an automated forecasting modeling process, with

manual intervention to improve models at certain key points and
automatic model iteration based on forecasting acuracy
• Created an automated process for forecast update across models
• Developed an interactive dashboard for forecasting view and
performance measurement

Technology

MS SQL, Server (For analysis and reporting data creation)

MS-Excel (for interactive reporting)
 Demand Forecasting
Demand Forecasting
Chapter 05
Chapter 05

S. No. Reference No. Particulars Slide From – To

1. Learning Objectives 82 – 82

2. Topic 1 Introduction 83 – 83

3. Topic 2 Steps in Demand Forecasting 84 – 84

4. Topic 3 Methods or Techniques of Demand 85 – 87

Forecasting

5. Topic 4 Summary 88 – 91

1– 5
 What is Forecasting?

¨ Process of predicting a future Sales will be

event Rs200 Million!

¨ Underlying basis of
all business decisions
¨ Production
¨ Inventory
¨ Personnel
¨ Facilities
Introduction

 Forecasting is an attempt to foresee the future by

examining the past.
 Virtually all types of national and international
organizations – Government, social and business – engage in
some type of demand forecasting.
 Demand forecasts may be passive or active.

1– 7
 Steps in Demand Forecasting

1. Identification of Objective
• Use of forecast
2. Nature of Product and Market
• E.g. Consumer good, perishable etc
3. Determinants of Demand
• Factors affecting the demand (Demography,
psychological etc)
4. Analysis of Factors
• Trend, cyclic, seasonal or random
5. Choice of Method
• Will depend on degree of accuracy, complexity
6. Testing Accuracy
• Testing statistical accuracy in a given forecast

1– 8
 Steps in Testing
Demand Forecasting
Accuracy

1– 9
Methods or Techniques of Demand Forecasting

Forecasting Methods
 Survey Methods
 Consumer Survey Method

 Collective Opinion Method

Delphi method

 Market Experiment Method

 Survey Methods
Surveys of consumer plans can be one of the important methods
of forecasting. The rationale for conducting such surveys is that
plans generally form the basis for future actions. By using this
method firms can ask consumers what and how much they are
planning to buy in the future.
Merits:
• A direct method of assessing info from primary source
• It is a simple and not based on past data
• Saves time and cost by conducting surveys base on a sample
• There is no bias or value judgment
Demerits
• There may be personal opinion or bias of the consumer
• It is difficult to ascertain number of customer who will buy
• The validity of this estimates is limited to about a year
• There can be sampling errors
 Collective Opinion Methods
Under this method salesmen are required to estimate expected
future demand of the product in their respective territories. The
rationale of this method is that salesmen being closest to the
customers are likely to have more intimate feel of the market
Merits
• The method is simple and does not involve statistical
techniques
• The forecast are based on first hand knowledge
• Is useful when forecasting new products
Demerits
• It is almost very subjective as personal opinion can influence
the forecast
• The usefulness is restricted to short term forecasting (1 yr)
• Salesmen may be unaware of the broader economic changes
likely to have an impact on the future demand
 Delphi Methods
It consists of an attempt to arrive at a consensus in an uncertain
area by questioning a group of experts until a the responses
appear to cover along a single line.

Merits
• Facilitates the maintenance of anonymity of the respondent’s
identity throughout the discussion
• This method renders it possible to pose the problem to the
expert at one time and have the responses.
• This techniques saves time

Demerits
• It is a tedious method
 Market Experiment Method
This method is popular in developed countries but not tried much
in India. Under this method the main determinants of the demand
of a product like price, advertising, product design, packaging
quality etc are identified. These factors are then varied separately
over different markets or over different time periods holding other
factors constant.
Merits
• This carefully carried out exercise can help the researcher to
come out with a demand function, indicating quantities that the
consumers will readily buy.
Demerits
• It is expensive
• These methods are risky as they give wrong signal to consumer,
dealers and competitors
Statistical Methods
 Time Series Analysis

 Graphical Method

 Semi Averages Method

 Moving Averages Method

 Least Squares Method

 Fitting Linear Trend

 Fitting Non Linear Trend

 Regression Analysis
 Time Series Components

Seasonal peaks Trend component

Demand for product or service

Actual demand
line

Average demand
over four years
Random
variation
Year Year Year Year
1 2 3 4
Trend Component

• Persistent, overall upward or downward

pattern
• Due to population, technology etc.
• Several years duration

Response

Mo., Qtr., Yr. © 1984-1994 T/Maker Co.

Seasonal Component

• Regular pattern of up & down fluctuations

• Due to weather, customs etc.
• Occurs within 1 year

Summer
Response
© 1984-1994 T/Maker Co.

Mo., Qtr.
Cyclical Component

• Repeating up & down movements

• Due to interactions of factors influencing
economy
• Usually 2-10 years duration

Cycle
Response

Mo., Qtr., Yr.

Random Component

 Erratic, unsystematic, ‘residual’ fluctuations

 Due to random variation or unforeseen
events
Union strike
Flood, Storm
 Short duration & non-repeating
 Semi-Average Method
According to this method, the data is divided into two
parts, preferable with the same number of years.
The average of the first and the second part are
calculated separately. These are then plotted.
A trend line is then obtained.
Year Sales (‘00) 3 yearly Semi-
Semi Totals Average
1997 20
1998 24 66 66/3 = 22
1999 22
2000 30 90 90/3 = 30
2001 28
2001 32
Semi-Average Method
Moving Average Method

• MA is a series of arithmetic means

• Used if little or no trend
• Used often for smoothing
– Provides overall impression of data over time
• Equation

MA   Demand in Previous n Periods

n
Moving Average Example

You’re manager of a museum store that sells historical replicas.

You want to forecast sales (000) for 2020 using a 3-period
moving average.

Year Sale (‘000)

2015 4
2016 6
2017 5
2018 3
2019 7
Moving Average Solution

Time Response Moving Moving

Yi Total Average
(‘000) (n=3) (n=3)
2015 4 NA NA
2016 6 NA NA
2017 5 NA NA
2018 3 4+6+5=15 15/3=5.0
2019 7 6+5+3=14 14/3=4.7
2020 NA 5+3+7=15 15/3=5.0
Simple Moving Average
 Simple Moving Average
• Forecast Ft is average of n previous observations or actuals Dt
:
1
Ft  1  ( D t  D t 1    D t  1  n )
n
1 t
Ft  1   Di
n i  t  1 n
• Note that the n past observations are equally weighted.
• Issues with moving average forecasts:
– All n past observations treated equally;
– Observations older than n are not included at all;
– Requires that n past observations be retained;
– Problem when 1000's of items are being forecast.
Weighted Moving Average Method

 Used when trend is present

 Older data usually less important
 Weights based on intuition
 Often lay between 0 & 1, & sum to 1.0
 Equation
Σ(Weight for period n) (Demand in period n)
WMA =
ΣWeights
 Weights Applied Period
3 Last month
2 Two months ago
1 Three months ago
6 Sum of weights

Actual 3-Month Weighted

Month Shed Sales Moving Average

January 10
February 12
March 13
April 16 [(3 x 13) + (2 x 12) + (10)]/6 = 12.2
May 19 [(3 x 16) + (2 x 13) + (12)]/6 = 14.3
June 23 [(3 x 19) + (2 x 16) + (13)]/6 = 17
July 26 [(3 x 23) + (2 x 19) + (16)]/6 = 20.5
Actual Demand,
Moving Average,
Weighted Moving Average
35 Weighted moving average

30
Actual sales
25
Sales Demand

20
15

10
Moving average
5
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
 Disadvantages of
Moving Average Methods

 Increasing n makes forecast less sensitive to

changes
 Do not forecast trend well
 Require much historical data
Linear Trend Projection
• Used for forecasting linear trend line
• Assumes relationship between response variable, Y, and
time, t, is a linear function

Yi = a + bt i

• Estimated by least squares method

– Minimizes sum of squared errors
Linear Regression Model

• Shows linear relationship between dependent &

explanatory variables
– Example: Sales & advertising (not time)

Y-intercept Slope

^
Yi = a + bti
Dependent
Independent (explanatory)
(response) variable
variable
Linear Regression Equations

Equation: Yˆi  a  bti

t y
Slope: b  2
t

y
Y-Intercept: a 
n
Interpretation of Coefficients

• Slope (b)
– Estimated Y changes by b for each 1 unit
increase in t
• If b = 2, then sales (Y) is expected to increase by 2 for
each 1 unit increase in advertising (t)
• Y-intercept (a)
– Average value of Y when t = 0
• If a = 4, then average sales (Y) is expected to be 4
when advertising (t) is 0
 Linear Trend Line

The following is the sales (in Rs Lakh) data of hand pumps. Find out what
can be the likely value of sales in the year 2003

Year Sales (Rs L)

1996 80
1998 90
1999 92
2000 83
2001 94
2002 99
2003 92
Finding a Trend Line

Year Sales t = year -1999 t2 ∑tY

( Rs L)
1996 80 -3 9 -240
1997 90 -2 4 -180
1998 92 -1 1 -92
1999 83 0 0 0
2000 94 1 1 94
2001 99 2 4 198
2002 92 3 9 276
Y= 630 t=0 t2 = 28 tY = 56
The Trend Line Equation

y
a  
630 = 90
n 7

t y
b 
56
 28 =2
2
t

Hence, the trend line for the sales of hand pumps is

Y = 90 + 2t
If t =4 (i.e year 2003) Y = 90 +2 (4) =Rs 98 L
 Linear Trend Method

This method of least squares is the most popular

and is one of most accurate methods of demand
forecasting particularly for long-term demand
projection because of is simplicity and
inexpensiveness.
However, it cannot explain satisfactorily market
fluctuations of trade cycle.
 Fitting Non Linear Trend
Demand having a linear function in time is an over
simplistic assumption. The demand function can be a
parabola, exponential logarithmic curves etc.
Least square method can be used to fit such cases of non
linear trends also.
When a new product is launched, in the initial phase the
demand will grow at a slow rate, followed by a fast rate, till
it reaches a saturation point.
 Non Linear Trend

industry has registered an exponential

country. Data is given for this industry for
1997 – 2001. Find the demand for the year

Year Sales (Rs L)

1997 1.6
1998 4.5
1999 13.8
2000 402
2001 125
 Non Linear Trend method
Solution
Data on sales shows that there is an exponential growth. So, we
have the following model
Y = ab t
i.e
log Y = log a + t log b
or y = A + Bt
where y = log Y, A = log a and B = log b
The least square equation will be
∑y = nA + B ∑t
∑ ty = A ∑t + B ∑t2
 Non Linear Trend
∑ y = nA + B ∑t ………………..equation 1
∑ ty = A ∑t + B ∑t2 ………………equation 2
t Y y=logY ty t2
1 1.6 0.2041 0.2041 1
2 4.5 0.6532 1.3064 4
3 13.8 1.1399 3.4197 9
4 40.2 1.6042 6.4168 16
5 125.0 2.0969 10.4845 25
∑t =15 ∑y =5.6983 ∑ty=21.8315 ∑t2= 55
Substituting the values in the above two least square equations we get
5.6983 = 5 A + 15 B ……..1 and 21.8315 = 15 A + 55 B ……….2
On solving these two simultaneous equations we have
A = log a = - 0.2814 and B = log b =0.4737
A =Antilog A = Antilog (-0.2814) = 0.5231 b=Antilog B = Antilog (0.4737) = 2.977
Non Liner Trend equation is Y =abt i.e Y = (0.5231) (2.977)t
 Non linear Trend Example
ndustry has registered an exponential
ountry. Data is given for this industry for
997 – 2001.
nd for the year 2005

d equation is
Y =abt
i.e Y = (0.5231) (2.977)t
t=9

Y = (0.5231) (2.977)9
= Rs 9607.142 Lakhs
nd for year 2005 is predicted as Rs 9607 L
 Regression Analysis

Regression analysis is the most popular method of forecasting

among economists. It is mathematical analysis of the average
relation between two or more variables in terms of the orginal
units of the data.
Here, the data analysis is based on the logic of economic theory.
The explanatory variable whose value is influenced or is to be
predicted is called the regressed or dependent variable and the
variable which influences the dependent or predicted value is
called the regressor or predictor variable
When the regression analysis is confined to the study of two
variables, it is called simple regression
 Regression Analysis
Merits
• This method is prescriptive as well as descriptive. Besides
generating demand forecast, it explains why the demand is,
what it is.
• It is an objective method using time series as well as cross-
section data
Demerits
• If some explanatory variable are not realistically chosen, they
trend to be misleading
• Forecast will be wrong to the extent that future relationship
deviates from the past average experience
• Regression results will be biased in case of auto-correlation, i.e.
when for one year demand is highly correlated with the
demand of preceding years
 LETS SUM UP

• Demand forecasting is a scientific exercise.

• The steps necessary for demand forecasting are identification
of objective, nature of product and market, determination of
demand, analysis of factors, choice of method and testing
accuracy
• Survey method are usually suitable for short-term forecasts
due to the nature of consumers’ intentions. New products
demand forecast also makes use of survey approach.
• Statistical methods are useful for long term forecasting for old
products and for larger levels of aggregations. They are based
on scientific ways of estimation
• Regression analysis is the most popular method of forecasting
among economists.
 To post a query relating to this course presentation please login
to Student Zone

1– 50
1– 51