FORECASTING

WHAT IS FORECASTING?

Process that can predict future sales or

demand by quantitative data or by
judgemental technique
These predictions give the basis of future
requirements of men ,material ,machine
and money
It is a projection based upon past data
and the art of human judgement
Types of Forecasting
FORECASTING

QUANTITATIVE METHOD For short or medium term QUALITATIVE Long term(2 to 5 years)
METHOD

Opinion Survey( Delphi

Time series(short term) Casual or Econometric
Technique)

Market trial
Simple Avg Method
Co-relation
Moving Avg Method Analysis
Market research
Weighted Moving Avg Linear
Exponential Smoothing regression
Analysis
Simple Average Method
Forecast sales for next period =Average sales for previous period

Period no Sales
1 7
2 5 7+5+9+8+5
Forecast sales for period no 6 =
3 9 5
4 8 = 6.8
5 5
 Moving Average Method
In this method the forecast is neither influenced by very old data nor does it soley reflect
the figure of the previous period

Year Period Sales 50+60+50+40

Forecast for 1988 period 1 = = 50
4
1987 1 50
2 60
3 50
4 40 60+50+40+50
Forecast for 1988 period 2 = = 55
4
1988 1 50
2 55
Weighted Moving Average
A weighted moving average allows any weights to be placed on each
element, providing of course , that the sum of all weights equal one

Period Sales

Month 1 100
Month 2 90
Month 3 105
Month 4 95
Month 5 110
Forecast ( Weights 40%,30%,20%,10% of most recent month)
𝐹5 = 0.4 ∗ 95 + 0.3 ∗ 105 + 0.2 ∗ 90 + 0.1 ∗ 100 = 97.5
𝐹6 = 0.4 ∗ 110 + 0.3 ∗ 95 + 0.2 ∗ 105 + 0.1 ∗ 90 = 102.5
Exponential Smoothening
We only require latest demand and old forecast
1. It’s a weighted average for fast observations
2. It assigns the highest weightage to the most recent observations
New forecast = 𝜶 𝒍𝒂𝒕𝒆𝒔𝒕 𝒔𝒂𝒍𝒆𝒔 𝒇𝒊𝒈𝒖𝒓𝒆𝒔 + 𝟏 − 𝜶 (𝑶𝒍𝒅 𝑭𝒐𝒓𝒆𝒄𝒂𝒔𝒕)
Where 𝛼 is known as the smoothing constant ; this constant gives equivalent of an
𝟐
N-period moving average can be calculated as follows 𝛼=
𝑵+𝟏
When N = 1 , 𝜶 = 1 and when N = infinite 𝜶=0

0≤𝛼≤1 STABLE
0 𝜶 1
RESPONSIVE
Exponential Smoothening
RESPONSIVENESS

If the demand of our product has a swinging pattern then it is known as a Responsive behavior with
respect to time

STABILITY

If the demand does not have a swinging pattern and almost remains constant with respect to time
then it is said to have a Stable behavior
Exponential Smoothening

General equation for Exponential smoothening

𝐹𝑡 = 𝛼𝐷𝑡−1 + 𝛼 1 − 𝛼 𝐷𝑡−2 + 𝛼(1 − 𝛼)2 𝐷𝑡−3 + ⋯

𝐹𝑡 = 𝛼𝐷𝑡−1 + 1 − 𝛼 [𝛼𝐷𝑡−2 + 𝛼 1 − 𝛼 𝐷𝑡−3 + ⋯ ∞
𝐹𝑡 = 𝛼𝐷𝑡−1 + (1 − 𝛼)𝐹𝑡−1
𝐹𝑡 = 𝐹𝑡−1 + 𝛼[𝐷𝑡−1 − 𝐹𝑡−1 ]
Forecasting Error
Forecast error helps to find the pattern in error and it regulates the future production.
The error should be minimum as for as possible & within a limit .These errors helps
to find a pattern or sequence to control activities of production system.

𝑬𝒓𝒓𝒐𝒓 = 𝑫𝒊 − 𝑭𝒊 = 𝒆𝒊 = ∆𝒊
Mean Absolute Deviation (MAD)
𝒏

෍ 𝑫𝒊 − 𝑭𝒊
𝒊=𝟏
n

It is calculated as the summation of absolute error for given period divided by the
number of periods
Forecasting Error
𝒏
Mean Forecast Error (MFE) or Bias
෍ 𝑫 𝒊 − 𝑭𝒊
𝒊=𝟏
n
It tells us the direction of error & here signs are considered . This error tells us
any chances of Over estimated & Under estimated . Positive value indicates Under
estimated forecasting and negative

Running sum Forecast Error (RSFE)

෍(𝑫𝒊 − 𝑭𝒊 )
𝒊=𝟏

Bias= RSFE / n
Forecasting Error
𝒏
Mean Square Error (MSE)
෍(𝑫𝒊 − 𝑭𝒊 )𝟐
𝒊=𝟏

Now a days it is the most used one . It is used for plotting control chart for Forecast
error .They magnify the error of larger magnitude.

𝒏
Mean Absolute Percentage Error (MAPE) 𝑫 𝒊 − 𝑭𝒊
෍ ∗ 𝟏𝟎𝟎
𝑫𝒊
𝒊=𝟏
n
It gives the forecasted error in proportion to actual demand in absolute terms . It
pulls error in perspective because there is difference between 10 errors out of 100 &
100 out of 1000
TRACKING SIGNAL = RSFE / MAD
Correlation Analysis
It is used in determining the degree of closeness or relationship between two variables
. It is an indication of the extent to which the knowledge of one variable is useful in
the prediction of other

The correlation between two variables x and y is given

ഥ ) . (𝒀 − 𝒀
෍(𝑿 − 𝑿 ഥ)

ഥ )𝟐 . ෍(𝒀 − 𝒀
෍(𝑿 − 𝑿 ഥ )𝟐
Linear Regression Analysis
This is mathematical technique of obtaining the line of best fit between the
dependent variable which is usually demand and some other independent variable on
which demand is dependent
𝑭𝒕 = 𝒂 + 𝒃𝒕 ( In diagram)
a, b = constants on which forecast depends
t = independent variable

In a simple regression analysis , the relationship

between the dependent variable ‘y’ and some
Independent variable ‘x’ can also be represented
as- Y = a + bx ………….(i)
Linear Regression Analysis
Now taking summation both side for period ‘n’ of the equation no (i)
෍ 𝒀 = 𝒂. 𝒏 + 𝒃 ෍ 𝑿 σ𝑌 − 𝑏σ𝑋
𝑎=
𝑛
Then multiplying by ෍ 𝑋 both sides
෍ 𝑋𝑌 = 𝑎 ෍ 𝑋 + 𝑏 ෍ 𝑋 2

σ𝑌 − 𝑏σ𝑋
෍ 𝑋𝑌 = ෍ 𝑋 + 𝑏 ෍ 𝑋2
𝑛
2
σ𝑋σ𝑌 − 𝑏 σ𝑋 + 𝑛𝑏 σ 𝑋 2
෍ 𝑋𝑌 =
𝑛
2

𝑛 ෍ 𝑋𝑌 − ෍ 𝑋 ෍ 𝑌 = 𝑏 𝑛 ෍ 𝑋 2 − ෍ 𝑋

𝒏 σ 𝑿𝒀 − σ 𝑿 σ 𝒀
𝒃=
𝒏 σ 𝑿𝟐 − σ 𝑿 𝟐
Qualitative methods
Judgemental
It’s based on the art of the human judgement like how well a human being can predict
a demand of product in future . This method doesn’t require past data sales figure.

Opinion Survey
In this method opinions are collected from the customers, relatives and distributor
regarding the demand pattern for a product . They give information why they buy a
particular product ,what cost they are willing to pay ,additional features required in
the product, their margin in profit etc.

Market Trial
This method basically used for a new product in this case it is advisable to introduce
the product in a limited population in the form of free samples . The response from the
population helps to project the demand from large population.

The cost of this method is high and it is used for low cost
consumables like Toothpaste, chocolates, cold drinks etc
Qualitative methods

Market Research
Here the work of survey is assigned to external marketing agencies &
the purpose of research is to collect information regarding the
consumption of a product , the details about various factor which
influence the demand like location ,Customer’s occupation ,
Customer’s income quantity and quality etc are scientifically related
to get the forecast
Delphi Technique
In this method a panel of experts are asked a series of questions in
which the response to one question is used to produce next question.
The information available to some experts are made available to
other experts . It is a step by step procedure in which opinion’s are
collected from the experts to arrive at a reliable forecast
 Least Square Method Forecasting
Method of least squares is a standard approach to the approximate solution of
overdetermined systems i.e sets of equations inwhich there are more equations in which
there are more equations than unknowns

“ Least square” means that the overall solution

Minimizes the sum of the squares of the errors made
in solving every single equation.
The best fit in the least-squares sense minimizes the
sum of squared residuals, a residual being the difference
between an observed value and the value provided by a
material
Least Square Method Forecasting
Least square problems fall into two categories – linear least square and non-linear least
square , depending upon whether or not the residuals are linear in all unknowns .

The linear list squares problem occurs in

statistical regression analysis ; it has closed
forum solution
The non linear problem has no closed solution
and is usually solved by iterative refinement ; at
each iteration the system is approximated by a
linear one, thus the core calculation is similar in both cases.
Trend line
The line of best fit to represent the future demand using past data is called trend line
TRACKING SIGNAL
It monitors the performance of the forecasting model and automatically indicates
whether the model needs to be revised .
3
It is used for preparation of tracking signal in which upper limit is taken as 𝑀𝑆𝐸

If the value of tracking signal goes beyond this limit then model needs to be revised.
OBJECTIVES OF FORECASTING
a) It determines the production rate

b) It suggests the need for changes in production methods and also for plant expansion

c) It forms basis for production budget, labor budget , material budget etc.

d) It helps deciding the extent of advertising and product distribution

Difference Between Forecasting & Prediction
FORECASTING
1. It involves the projection of the
part into the project.
2. It is more scientific and more
objective
3. Error analysis is possible
4. It is reproducible
5. Free from personal bias
PREDICTION
1. It involves judgement in
management after taking all
available information into
account
2. It is more intuitive and
Subjective
3. No error analysis
4. It is non-reproducible
5. Governed by personal bias and
preferences
IMPORTANCE OF FORECASTING
1. Forecasting provides relevant and reliable information
about the past and present events and likely future events
.This is necessary for sound planning.
2. It gives confidence to managers to take important decisions
3. It is the basis for making the Planning Premises.
4. It keeps the managers active and alert to face the
challenge of future events and for the change in
environment.
THANK YOU