FUNDAMENTALS

OF
FORECASTING
 What is
Forecasting?
Forecasting is a technique for using past experiences to
project expectations for the future. Forecasting is not really
a prediction, but a structured projection of past knowledge
Some are long-range, aggregated models
used for long-range planning such as overall capacity
needs, developing strategic plans, and making long-term
strategic purchasing decisions. Others are short-range
forecasts for particular product demand.
 The Need for
Forecasting?
Accounting Cost/profit estimates
Forecasts
Finance Cash flow and funding
affect
decisions & Human Resources Hiring/recruiting/training

activities Marketing Pricing, promotion, strategy

throughout an
MIS IT/IS systems, services
organization
Operations Schedules, MRP, workloads

Product/Service
Design New products and services
Elements of a Good
Forecast

Timely

Reliable Accurate

Easy
Meaningful Written
to
Use
Forecasting
Process
 Timeline-Based
Classifi cations
Depending on the time horizon forecasting can be

classified as:
• Short range forecasting: up to 1 year
• Medium range: up to 3 years
• Long range: more than 3 years
 Forecasting
Techniques/Models

Qualitative Techniques Quantitative Techniques

Market Surveys Causal Time Series

Simple Moving
Life cycle Input Output Model
Averages
Analogy
Weighted Moving
Input Output Model
Delphi Averages
Technique
Simple Exponential
Simulation Model
Smoothing
Informed
Judgement
Regression Model Regression
 Model
Diff erences
• Qualitative – incorporates judgmental & subjective
factors into forecast.
• Time-Series – attempts to predict the future by
using historical data.
• Causal – incorporates factors that may influence the
quantity being forecasted into the model
 Qualitative: Market Survey
Method :
• Structured questionnaires are submitted to potential customers in the
market asking for opinions about products/ potential products
• The objective is to get an understanding of the possible demand for
products or services.
Advantages :
• If structured well and directed to the right target audience they can be
quite effective, especially for the short term.
Drawbacks :
• Expensive and
• Time-consuming
 Qualitative: Delphi
Method :
Technique
• Choose a facilitator: Required to coordinate between the panelists.
• Identify experts: Find multiple experts in the field of interest -could
be from within or outside the organisation.
• Define the research question to exact terms.
• Carry out three response and feedback rounds: Start by asking
for anonymous answers to the specific question. Then, present the
information to the experts and allow them to adapt the response in
light of the new information. Process is repeated normally around 3
times
• Distill & Summarise the final findings
 Qualitative: Delphi
Advantages: Technique
• Anonymity: group members can express their opinions freely without
being affected by peer pressure.
• Iterative feedback: Feedback Rounds give a bird’s-eye view of what
the rest of the panel members are thinking. This gives the experts an
insight into how they might adapt their response.
• Using experts: Use of experts vs random sample selection gives
better accuracy of the forecast.
Drawbacks:
• Expensive to get a panel of experts
• Time Consuming
• If panelists have divergent views possibility of no consensus.
 Qualitative: Delphi
Trivia :
Technique
• First used by the US as a technique for developing military strategies
and weapons in the Cold War Era.

• Named after the Greek The Oracle of Delphi where people would ask
the high priestess Pythia for guidance & wisdom.

• Difference between Brainstorming and Delphi Technique -

Brainstorming is an open face-to-face discussions where participants
share ideas freely. The Delphi technique uses anonymous
questionnaires to gather opinions, and allows experts to express their
true views.
Qualitative: Life Cycle
Analogy
It is based on the fact that most products
and services have a fairly well-defined
life cycle.

• What is the time frame? How long will

growth and maturity last?
• How rapid will the growth be? How
rapid will the decline be?
• How large will the overall demand be,
especially during the mature phase?
 Qualitative: Life Cycle
Analogy
Method:
• Link the demand for a new product/service to a similar one in the past.
It assumes the life-cycle for the new product/service will be roughly the
same as for the old product or service it is replacing.
Advantages :
• Effective if the new product/ service is essentially replacing another in
the market, targeted to the same population.
Drawbacks
• Change in consumer preferences, other offerings by competetitors can
affect the forecast
 Qualitative: Informed
Judgement
Method:
• Ask members of the team for their projection or forecast of the item in
question such as demand/sale. Individual estimates are combined into
the overall projections/forecasts .
Advantages :
• Tends to use employee experience.
Drawbacks
• Under or Over estimation of results due to individual outlooks can skew
the results
• Often used method but due to subjective nature, can give wrong results
 Quantitative: Causal
• Based on the conceptTechniques
that one measurable variable "causes" the
other to change in a predictable fashion.
• The measured variable that causes the other to change is often called
a "leading indicator."
• A leading indicator gives insight into other market knowledge. Eg
Leading indicators in economics (inflation, GDP etc) determine govt
policy matters and provide a reference to conglomerates on potential
demand patterns
• Methods are usually used for markets or industries & not for a product
.
• Time-consuming
• Expensive
 Causal: Input-Output Model
• It examines the flow of goods and services throughout the entire
economy.
• Needs a substantial quantity of data for the forecast to be accurate.
• Used to project needs for entire markets rather than for specific
products.
• Expensive.
• Time-consuming to develop.
 Causal: Econometric Model
• These models involve a statistical analysis of various
sectors of the economy.
• Their use is similar to the input-output models.
 Causal: Simulation Model
Method:
• Getting relevant data and plotting it to simulating patterns in sectors
of the economy.

Advantages
• Growing in popularity with better software & simulation models.
• Used for market analysis as well as individual products.
• Fast & economical once data has been collected.

Drawbacks
• Data collection on which the simulation model is based is expensive
and time-consuming.
 Causal: Regression Model
A statistical method to develop a defined analytic relationship
between two or more variables.
 Quantitative: Time Series
• Based on the assumption is that past demand follows some pattern
which is analyzed to develop projections of future demand,
assuming the pattern continues in roughly the same manner.

• Implies that the only real independent variable in the time series
forecast is time.

• Based on internal data (sales), they are also called intrinsic

forecasts

• Most commonly used methods for forcasting demand as data is

available inhouse for analysis.

• Important to understand trends in demand patterns.

 Quantitative: Time Series
Used when a manager wants to systematically analyze historical data
for forecasting

How to do time series analysis

• Plot demand data on a time scale
• Study the plots
• Look for consistent shapes or patterns.
• Analyse the patterns to forecast future demands.
Time Series: Demand Patterns
Time Series: Demand
Patterns/Trends
 Time Series: Simple Moving
Method
Averages
• Average of demands occurring in several of the most recent
periods

• Most recent periods are added up and older ones dropped to

keep calculations current

• Once the number of past periods to be used is selected, it is

held constant
Time Series: Simple Moving
Averages

Where:
F is the forecast
t is the current time period, meaning
Ft is the forecast for the current time
period
At is the actual demand in period t,
and
n is the number of periods being
used.
 Time Series: Simple Moving
Averages

• The demand for iPhones in Reliance Digital for the

past 5 weeks was as follows. Compute a three-period
simple moving average forecast for demand in week
6.
80 85 85 90 95
 Time Series: Weighted Moving
Averages
Method
Similar to Simple Moving Average but
• All periods are not equally weighted
• A weight is assigned to each past demand point.
• Weights can vary.
• More infl uence can be given to some data points,
(normally the most recent demand point).
• The sum of the weights add upto 1
Time Series: Weighted Moving
Averages
Simple Moving Weighted Moving
Actual
Period Weight Avg Demand Avg Demand
Demand
Forecast Forecast
1 24 0.2
2 26 0.3
3 22 0.5
4 25 24.00 23.60
5 19 24.33 24.30
6 31 22.00 21.40
7 26 25.00 26.20
8 18 25.33 26.10
9 29 25.00 23.00
10 24 24.33 25.10
11 23.67 24.30
 Time Series: Weighted Moving
Averages
• The demand for iPhones in Reliance Digital for the
past 5 weeks was as follows. Compute a three-
period weighted moving average forecast for
demand in week 6.
80 85 85 90 95
• Using a weight of .50 for the most recent
period, .30 for the next most recent, .20 for the
next.

• If the actual demand for week 6 is 93, forecast

demand for week 7 using the same weights.
Time Series: Simple & Weighted
Moving Averages
Simple
Wghted
Moving
Actual Moving Avg
Period Weight Avg
Demand Demand
Demand
Forecast
Forecast

1 80 0.2

2 85 0.3

3 85 0.5

4 90 83.33 84.00

5 95 86.67 87.50

6 93 90.00 91.50
 Time Series: Simple Exponential
Smoothing
It assumes a time series has no trend or seasonality.
It was fi rst developed by Brown in 1956

This method seeks to smoothen the random

fl uctuations in the demand pattern.

Alpha is used as a smoothing parameter. (0,1)

One of the most powerful and effi cient forecasting

methods for level time series.
 Time Series: Simple Exponential
Smoothing

• This implies that you must start the process

using another forecasting method, after which
the forecast from that method can be used as
the initial Ft - I
 Time Series: Simple Exponential
Smoothing
Simple
Exponential
Moving
Actual Wghted Moving Avg Smoothing with
Period Weight Avg
Demand Demand Forecast Alpha=.02 & yr 3
Demand
Forecast
forecast is 25
In this table we
1 24 0.2 have used a simple
2 26 0.3 moving average
3 22 0.5 for the fi rst two
4 25 24.00 23.60 24.4
periods to develop
an initial forecast
5 19 24.33 24.30 24.5
of 25 units for
6 31 22.00 21.40 23.4
period 3, after
7 26 25.00 26.20 24.9
which exponential
8 18 25.33 26.10 25.1 smoothing can be
9 29 25.00 23.00 23.7 use
10 24 24.33 25.10 24.8

11 23.67 24.30
 Time Series: Regression

• Demand History is placed in a statistical

package & the regression value is
calculated.
• A particular value for regression is to
determine trend line equations.
• Often referred to the line of best fi t
 Why Is Forecasting Important

• Scheduling
• Best use of existing conversion system
• 3 month, 6 month / 1 year forecasts facilitate to
schedule jobs
• Scheduling of work force levels & production rates

• Controlling
• Control over inventory, production, labor & over
all costs
 Forecast Errors
Numeric diff erence of forecasted demand and actual demand.

Answers - How incorrect is the forecast?

• Important in planning and control (Buff er inventory, buff er
capacity etc may be needed to be planned to
accommodate actual demand that diff ers from that
forecasted.

Methods of measuring forecast error

• Mean Forecast Error (Bias)

• MAD : Mean absolute deviation

 Forecast Errors = Mean Forecast
Error
• The mathematical average forecast error over a
specifi ed time period

• (At - Ft) represents the diff erence between the

forecast and the actual demand for any given
time period, also called the forecast error.

• The MFE involves adding all individual forecast

errors and dividing by the total number of errors.
 Forecast Errors = Mean Forecast
Error
• If positive, it implies the actual demand was
larger than the forecast. Another way of putting
that is that the forecasting method was biased
on the low side.

• If negative, of course, it means the forecasts were

larger than the demand on average, implying the
forecasting method was biased on the high side.

• For this reason, MFE is often referred as forecast

bias.
Forecast Errors = Mean Forecast
Error
 Forecast Errors = Mean Absolute
Deviation
• The average of the mathematical absolute
deviations of the forecast errors (deviations).

• The average forecast error in this case is always

positive
Forecast Errors = Mean Absolute
Deviation
THANK YOU