3. Objectives ?
• What is sales forecasting?
• Why doing sales forecasting?
• How Sales Forecasting Works
• What methods are used?
• How to use these methods?
Just like a ship's captain, it's up to sales
forecasting professionals to keep businesses on
course.
4/8/2013 Eng. Magdy Abdelsattar 3
5. First day: Agenda
1. introduction to sales forecasting
oWhat?
oWhy?
oHow?
2. Qualitative methods
oDelphi
oExpert Judgment
oScenario Writing
oIntuitive approach
oGroup Work
3. Quantitative methods
I. Time Series Methods
oTrend Component
oCyclical Component
oSeasonal Component
oIrregular Component
oExcel work
4/8/2013 Eng. Magdy Abdelsattar 5
6. Agenda
Second day:
Quantitative methods
I. Smoothing Methods
oMoving Average
oWeighted Moving Average
oExponential Smoothing
oExcel work
II. Casual Section
oRegression analysis casual method
oRegression analysis with time series
oExcel work
III. Trend and Seasonal
oMultiplicative Model
oSeasonal Indexes
oDeseasonalized the Time Series
oUsing DTS to Identify trends
oSeasonal Adjustment
oExcel Work
4/8/2013 Eng. Magdy Abdelsattar 6
8. What is sales forecasting?
4/8/2013 Eng. Magdy Abdelsattar 8
9. Sales forecasting.
A forecast is simply a prediction of what will happen in the future. Managers
must learn to accept the fact that, regardless of the technique used, they will
not be able to develop perfect forecasts.
Sales forecasting is a difficult area of management. Most of us believe we are
good at forecasting. However, forecasts made usually turn out to be wrong!
Marketers argue about whether sales forecasting is a science or an art. The
short answer is that it is a bit of both.
Most companies can forecast total demand for all products, as a group, with
errors of less than 5%. However, forecasting demand for individual products
may results in significantly higher errors.
With sales forecasting, companies can plan for future inventory on a monthly basis.
4/8/2013 Eng. Magdy Abdelsattar 9
10. Key terms in sales forecasting
Market demand:
for a product or service is the estimated total sales volume in a market (or industry) for
a specific time period in a defined marketing environment, under a defined marketing
program or expenditure. Market demand is a function associated with varying levels of
industry marketing expenditure.
Market forecast (market size):
is the expected market (industry) demand at one level of industry marketing
expenditure.
Market potential:
is the maximum market (industry) demand, resulting from a very high level of industry
marketing expenditure, where further increases in expenditure would have little effect
on increase in demand.
Company demand:
is the company’s estimated share of market demand for a product or service at
alternative levels of the company marketing efforts (or expenditures) in a specific time
period.
4/8/2013 Eng. Magdy Abdelsattar 10
11. Key terms in sales forecasting
Sales potential:
is the maximum estimated company sales of a product or service, based
on maximum share (or percentage) of market potential expected by the
company.
Sales forecast:
is the estimated company sales of a product or service, based on a chosen
(or proposed) marketing expenditure plan, for a specific time period, in a
assumed marketing environment
Sales budget:
is the estimate of expected sales volume in units or revenues from the
company’s products and services, and the selling expenses. It is set
slightly lower than the company sales forecast, to avoid excessive risks
4/8/2013 Eng. Magdy Abdelsattar 11
12. Type of forecasting
There are two major types of forecasting:
Macro forecasting:
is concerned with forecasting markets in total. This is about determining the
existing level of Market Demand and considering what will happen to
market demand in the future.
Micro forecasting:
is concerned with detailed unit sales forecasts. This is about determining a
product’s market share in a particular industry and considering what will
happen to that market share in the future.
4/8/2013 Eng. Magdy Abdelsattar 12
13. types of forecasting Information
Sales forecasts can be based on three types of information:
What customers say about their intentions to continue buying products in
the industry
What customers are actually doing in the market
What customers have done in the past in the market
4/8/2013 Eng. Magdy Abdelsattar 13
14. Sales forecasts also rely on obtaining information on existing market demand:
As a starting point for estimating market demand, a company needs to know the actual
industry sales taking place in the market. This involves identifying its competitors and
estimating their sales.
An industry trade association will often collect and publish (sometime only to members)
total industry sales, although rarely listing individual company sales separately. By using
this information, each company can evaluate its performance against the whole market.
4/8/2013 Eng. Magdy Abdelsattar 14
15. Factors affecting Forecasting
External Factors
o Relative state of the economy
o Direct and indirect competition
o Styles or fashions
o Consumer earnings
o Population changes
o Weather
4/8/2013 Eng. Magdy Abdelsattar 15
16. Factors affecting Forecasting
Internal Factors
o Labour problems
o Inventory shortages
o Working capital shortage
o Price changes
o Change in distribution method
o Production capability shortage
o New product lines
4/8/2013 Eng. Magdy Abdelsattar 16
17. forecasting problems.
The selection of which type of forecasting to use depends on several factors:
The degree of accuracy required –
if the decisions that are to be made on the basis of the sales forecast have high
risks attached to them, then it stands to reason that the forecast should be
prepared as accurately as possible although this involves more cost.
The availability of data and information –
in some markets there is a wealth of available sales information (e.g. clothing
retail, food retailing, holidays); in others it is hard to find reliable, up-to-date
information.
4/8/2013 Eng. Magdy Abdelsattar 17
18. forecasting problems.
The time horizon that the sales forecast is intended to cover.
For example, are we forecasting next weeks’ sales, or are we trying to forecast
what will happen to the overall size of the market in the next five years?
The position of the products in its life cycle.
For example, for products at the “introductory” stage of the product life
cycle, less sales data and information may be available than for products at the
“maturity” stage when time series can be a useful forecasting method.
4/8/2013 Eng. Magdy Abdelsattar 18
19. How to Improve Forecasting Accuracy?
Sales forecasting is an important & difficult task
Following guidelines may help in improving its accuracy
oUse multiple (2/3) forecasting methods.
oSelect suitable forecasting methods, based on application, cost, and
available time.
oUse few independent variables / factors, based on discussions with
salespeople & customers.
oEstablish a range of sales forecasts – minimum, intermediate, and
maximum.
oUse computer software forecasting packages.
4/8/2013 Eng. Magdy Abdelsattar 19
20. Forecasting Approaches
• Two basic approaches:
• Top-down or Break-down approach
• Bottom-up or Build-up approach
• Some companies use both approaches to
increase their confidence in the forecast
4/8/2013 Eng. Magdy Abdelsattar 20
21. Steps followed in Top-down / Break-
down Approach
o Forecast relevant external environmental factors
o Estimate industry sales or market potential
o Calculate company sales potential = market potential
x company share
o Decide company sales forecast (lower than company
sales potential because sales potential is maximum
estimated sales, without any constraints)
4/8/2013 Eng. Magdy Abdelsattar 21
22. Steps followed in Bottom-up / Build-up
Approach
o Salespersons estimate sales expected from their
customers.
o Area/Branch managers combine sales forecasts
received from salespersons.
o Regional/Zonal managers combine sales forecasts
received from area/branch managers.
o Sales/marketing head combines sales forecasts.
received from regional/zonal managers into company.
sales forecast, which is presented to CEO for discussion
and approval.
4/8/2013 Eng. Magdy Abdelsattar 22
23. Why we need sales forecasting?
1. Businesses are forced to look well ahead in order to plan their
investments, launch new products, decide when to close or withdraw
products and so on.
2. The sales forecasting process is a critical one for most businesses.
3. Key decisions that are derived from a sales forecast include:
oEmployment levels required
oPromotional mix
oInvestment in production capacity
oPlant Capacity & Projected Utilization
oAvailability of Raw Materials
oWorking Capital Requirements
oCapital Expenditure
oReturn on Investment Sales forecasting helps retailers decide how
many styles of a product to stock.
4/8/2013 Eng. Magdy Abdelsattar 23
24. Why doing sales forecasting?
A sales forecast is a projection of the coming year's sales revenue based on information
collected from the individual members of the sales team and sales management.
Distribution Process:
Sales forecasts identify not only the volume of sales, but where those sales are projected
to come from.
By using sales forecasting to do demand planning, the company can determine where
new distribution outlets are needed and decide on the best way to expand its product
network.
Manufacturing:
The level of manufacturing for any company is determined by the forecast of product
demand. In order to properly plan the acquisition of materials, schedule manufacturing
and determine the adequate personnel to meet that schedule.
Revised sales forecasts during the course of the year are also helpful in keeping
manufacturing up to date on needs and trends.
4/8/2013 Eng. Magdy Abdelsattar 24
25. Why doing sales forecasting?
Logistics:
An increase or decrease in sales forecasting is going to affect the logistics portion of
demand planning.
Sales forecasting is used to determine whether or not new logistics agreements need to
be negotiated with carriers and if the company needs to revise shipping policies.
Sales Force Expansion:
A growing company is going to experience a rise in demand that needs to be addressed
with an increased sales force.
Some potential sales force changes include creating new sales territories, splitting
existing territories into more sales regions and adding new sales representatives to
attend to those regions, and hiring more sales professionals to take care of an
expanding client demand.
4/8/2013 Eng. Magdy Abdelsattar 25
27. How Sales Forecasting Works
Collect and
analysis
data
Calculate
sales
forecast
Determine
forecasting
methods
It is all about determining future market demand, through an analysis of the current market
and past sales data
4/8/2013 Eng. Magdy Abdelsattar 27
28. What are the steps for that?
“The forecast”
Step 7 Validate and implement results
Step 6 Make the forecast
Step 5 Obtain, clean and analyze data
Step 4 Select a forecasting technique
Step 2 Select the items to be forecasted
Step 1 Determine purpose of forecast
4/8/2013 Eng. Magdy Abdelsattar 28
29. Preparing a Sales Forecast
• Very few products or services lend themselves to easy forecasting .
• In most markets, total demand and company demand are not stable – which
makes good sales forecasting a critical success factor.
Prepare a Prepare an industry Prepare a company
macroeconomic sales forecast sales forecast
forecast
• what will happen to • what will happen to • based on what
overall economic overall sales in an management expect
activity in the industry based on to happen to the
relevant economies the issues that company’s market
in which a product is influence the share
to be sold. macroeconomic
forecast;
4/8/2013 Eng. Magdy Abdelsattar 29
30. How should we pick our forecasting
model?
1. Data availability
2. Time horizon for the forecast
3. Required accuracy
4. Required Resources
4/8/2013 Eng. Magdy Abdelsattar 30
31. Some Important Questions
• What is the purpose of the forecast?
• Which systems will use the forecast?
• How important is the past in estimating the future?
Answers will help determine time
horizons, techniques, and level of detail for the forecast.
4/8/2013 Eng. Magdy Abdelsattar 31
32. Forecasting Methods
Forecasting is the process in
business of determining what
the business market that you
are engaged in looks like
demographically
It can also involve attempting to
predict the movements of
the existing market going
forward so market strategies
and business plans can be
developed to anticipate and
meet the changing demands
4/8/2013 Eng. Magdy Abdelsattar 32
34. •Qualitative Forecasting Methods:
Qualitative forecasting methods attempt to use actual data to determine a qualitative or
actual market trend toward a certain position or function in the market. These methods
involve looking at non-numerical data. Qualitative forecasting methods are not as
effective as quantitative methods,
•Quantitative Forecasting Methods:
In general, quantitative methods use numbers -- sales numbers
Explanatory Methods
Explanatory forecasting methods use data to attempt to explain trends and to forecast
future market direction based on existing data. Explanatory methods involve looking at
market activity to explain how and why trends occurred, not just to predict what will
occur.
Time-series Methods
Time-series methods are used only with historical data to predict future performance.
4/8/2013 Eng. Magdy Abdelsattar 34
36. Delphi Method
•originally developed by a research group at the Rand
Cooperation, attempts to develop forecasts through “group consensus”.
•The members of a panel of experts-all of whom are physically
separated from and unknown to each other-are asked to respond to a
series of questionnaires.
•The response of the first questionnaire are tabulated and used to
prepare a second questionnaire that contains information provided.
•This process continues until the coordinator feels that some degree of
consensus has been reached.
•The objective is to produce a relatively narrow spread of opinions
within which the majority of experts concur.
4/8/2013 Eng. Magdy Abdelsattar 36
37. Expert Judgment
•Often are based on the judgment of a single expert or
represent the consensus of a group of experts.
•In doing so, the experts individually consider
information that they believe will influence the market
, then they combine their conclusions into a forecast.
•No formal model is used, and no tow experts are likely
to consider the same information in the same way.
4/8/2013 Eng. Magdy Abdelsattar 37
38. Scenario Writing
•Scenario writing consists of developing a
conceptual scenario of the future based on a
well-defined set of assumptions.
•The job of the decision maker is to decide
how likely each scenario is and then to make
decision accordingly.
4/8/2013 Eng. Magdy Abdelsattar 38
39. Intuitive Approaches
•Subjective, or intuitive qualitative approaches, are
based on the ability of the human mind to process
information that, in most cases, is difficult to quantify.
•These techniques are often used in group
work, wherein a committee or panel seeks to develop
new idea or solve complex problem through a series of
“brainstorming sessions”.
4/8/2013 Eng. Magdy Abdelsattar 39
40. Group Activity
In teams try to implement
qualitative analysis to demonstrate
the concepts
4/8/2013 Eng. Magdy Abdelsattar 40
42. Quantitative Methods
Decomposition
Process:
includes breaking down the company’s previous periods’ sales data into
components like trend, cycle, seasonal, and erratic events. These components are
recombined to produce sales forecast
Advantages:
Conceptually sound, fair to good accuracy, low cost, less time
Disadvantages:
complex statistical method, historical data needed, used for short-term
forecasting only
4/8/2013 Eng. Magdy Abdelsattar 42
43. Time-Series
Using time series analysis to prepare an
effective sales forecast requires
management to:
•Smooth out the erratic factors (e.g.
by using a moving average)
•Adjust for seasonal variation
•Identify and estimate the effect of
specific marketing responses
Time series analysis are accurate for short term
and medium term forecasts and more so when
demand is stable or follows the past behavior.
4/8/2013 Eng. Magdy Abdelsattar 43
44. Time-Series
•Sales History:
Sales history is an important tool in
forecasting. It's the basis for
inventory, staffing and business resource
planning.
Knowing previous years' sales allows the
establishment of a baseline, or starting
point, for setting goals.
Sales history, analyzed with knowledge of
the market, customers, industry and
products, is the main indicator of future
sales opportunities.
4/8/2013 Eng. Magdy Abdelsattar 44
45. Time-Series
Open-model time-series techniques involve analyzing sales history data for
patterns to use in sales forecasting. These are patterns in level, trends, Cyclical
and seasonality, combined with "noise."
Level:
is the sales history without trends.
Trends Component:
are increases or decreases in sales that continue year after year.
Cyclical Component:
Any frequent sequence of sales above and below the trend line lasting more than
one year. Sales are often effected by swings in general economic activity as
consumers have more or less disposable income available
4/8/2013 Eng. Magdy Abdelsattar 45
46. Time-Series
Seasonality Component:
is a pattern of sales of particular items at particular times of the year,
Noise (Irregular Component):
involves random effects in sales that don't have a repeatable pattern in
previous sales.
Analyzing sales history trends and reasons for changes in sales enables sales
personnel to produce more accurate forecasts.
4/8/2013 Eng. Magdy Abdelsattar 46
48. Naïve Forecasts
The forecast for any period equals the previous period’s actual value.
oSimple to use
oVirtually no cost
oQuick and easy to prepare
oData analysis is nonexistent
oEasily understandable
oCannot provide high accuracy
Stable time series data
F(t) = A(t-1)
Seasonal variations
F(t) = A(t-n)
Data with trends
F(t) = A(t-1) + (A(t-1) – A(t-2))
4/8/2013 Eng. Magdy Abdelsattar 48
49. Naïve Forecasts
Naïve / Ratio method
Assumes:
what happened in the immediate past will happen in immediate future
Simple formula used:
Actual sales of this year
Sales forecast for next year Actual sales of this year
Actual sales of last year
Advantages:
simple to calculate, low cost, less time, accuracy good for short-term
forecasting
Disadvantages:
less accurate if past sales fluctuate
4/8/2013 Eng. Magdy Abdelsattar 49
50. Using Excel
1. Using Time Series
4/8/2013 Eng. Magdy Abdelsattar 50
51. Second day
Good Morning
4/8/2013 Eng. Magdy Abdelsattar 51
53. Day one Recap
oDefinitions
oWHY DO WE FORECAST ?
oScope of Forecasting
oAdvantages & Disadvantages
oForecasting Time Horizon
oSources of Data
oTypes of Forecasting
4/8/2013 Eng. Magdy Abdelsattar 53
54. Definitions
oIt is estimating the future demand for products & services & the resources
necessary to produce these outputs
or
oForecasts is the essence of management . Its techniques are used in every
types of organization may be it government or private, production or service &
social or religious
Forecasts are critical inputs to business plans, & budgets .
Finance – predict cash flows & capital requirements.
Human Resource – To anticipate hiring & training needs.
Operations – forecasts to plan output levels, purchase, output
schedules, inventory , capacity planning
4/8/2013 Eng. Magdy Abdelsattar 54
55. Why ?
1) Short term fluctuations in Demand
2) Better materials management – Organizations can benefit from better
materials management, & ensure materials are available in time.
3) Manpower Decisions – Hiring or layoff
4) Basis for Planning & scheduling- planning & scheduling can be done
effectively
5) Strategic Decisions – Useful for Long range strategic decision making. This
includes planning for product line decisions, new products etc.
4/8/2013 Eng. Magdy Abdelsattar 55
56. Advantages & Disadvantages
Advantages:
oHelps in Effective planning
oHelps in better co-ordination
oAchieves co-operation in Enterprises
oEffective Control
Disadvantages:
oBased on assumptions
oBased on past data
oNot Full Proof
oInadequate data
4/8/2013 Eng. Magdy Abdelsattar 56
57. Sources of Data
Sales Force Estimate:
One of the most valuable sources of data & quality of data that is available is the sales force that operates in
the field. Since sales force spans the entire geographic range of operation they have access to data pertaining to
consumption, changing patterns , market growth
Points of Sales ( POS ) Data Systems:
sort of information technology . In supermarket if you buy Surf excel , at check counter when sales person
swipes pack through POS system, the data is captured & transmitted to the relevant database for the company
to analyze
Forecasts from supply Chain Partners:
Obtaining POS data from distributors & suppliers
Trade / Industry Association Journals:
These journals provide research data on the sector in which the organization is operating ( Automobile sector )
4/8/2013 Eng. Magdy Abdelsattar 57
58. Sources of Data
B2B Portals / Market Places :
Another source of data in the era of www is the existence of industry portals & B2B
market places. For agricultural www.industryindiaagronert.com , for small &
medium sector enterprises www.sme.in
Economic Surveys & Indicators :
Studied conducted by research organizations on macroeconomic trends are good
indicators of emerging trends in the consumption patterns of several classes of
goods & services .e.g. Centre for monitoring Indian Economy (CMIE), Consensus
Economics
Subjective Knowledge :
Long-term Forecasts enable strategic decision making. Senior Managers, subject
experts are vital source of data.
4/8/2013 Eng. Magdy Abdelsattar 58
60. Moving Average
Moving average
• The moving average model uses the last t periods in order to predict demand in
period t+1.
• There can be two types of moving average models: 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.
4/8/2013 Eng. Magdy Abdelsattar 60
61. Moving Average
Moving averages
Procedure:
is to calculate the average company sales for previous years
Moving averages name is due to dropping sales in the oldest period and replacing it
by sales in the newest period
Advantages:
simple and easy to calculate, low cost, less time, good accuracy for short term and
stable conditions
Disadvantages:
can not predict downturn / upturn, not used for unstable market conditions and
long-term forecasts
4/8/2013 Eng. Magdy Abdelsattar 61
62. simple moving average
In the simple moving average models 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.
4/8/2013 Eng. Magdy Abdelsattar 62
63. Example:
Coca-Cola sells (among other stuff) bottled water
Month Bottles
Jan 1,325 What will the sales be
Feb 1,353 for July?
Mar 1,305
Apr 1,275
May 1,210
Jun 1,195
Jul ?
4/8/2013 Eng. Magdy Abdelsattar 63
64. 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
1400
1350
1300 5-month
MA forecast
1250
1200 3-month
1150 MA forecast
1100
1050
1000
0 1 2 3 4 5 6 7 8
4/8/2013 Eng. Magdy Abdelsattar 64
65. Stability versus responsiveness in moving averages
What do we observe?
950
900
850
800
Demand
750
700 3-Week
650
6-Week
600
550
500
1 2 3 4 5 6 7 8 9 10 11 12
5-month average smoothes data more;
3-month average more responsive
4/8/2013 Eng. Magdy Abdelsattar 65
66. weighted moving average
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
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.
w :is the importance (weight) we give to each period
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.
4/8/2013 Eng. Magdy Abdelsattar 66
67. Example:
Coca-Cola sells (among other stuff) bottled water
Month Bottles
Jan 1,325 What will the sales be
Feb 1,353 for July?
Mar 1,305
Apr 1,275
May 1,210
Jun 1,195
Jul ?
4/8/2013 Eng. Magdy Abdelsattar 67
68. 6-month simple moving average…
AJun + AMay + AApr + AMar + AFeb + AJan
FJul = = 1,277
6
In other words, because we used equal weights, a
slight downward trend that actually exists is not
observed…
4/8/2013 Eng. Magdy Abdelsattar 68
69. What if we use a weighted moving average?
Make the weights for the last three months more than the first
three months…
6-month WMA WMA WMA
SMA 40% / 60% 30% / 70% 20% / 80%
July
1,277 1,267 1,257 1,247
Forecast
The higher the importance we give to recent data, the more we
pick up the declining trend in our forecast.
4/8/2013 Eng. Magdy Abdelsattar 69
70. How do we choose weights?
1. Depending on the importance that we feel past data has
2. Depending on known seasonality (weights of past data
can also be zero).
WMA is better than SMA
because of the ability to
vary the weights!
4/8/2013 Eng. Magdy Abdelsattar 70
71. Exponential Smoothing (ES)
Main idea:
The prediction of the future depends mostly on the most recent
observation, and on the error for the latest forecast.
Smoothing Denotes the
constant alpha importance of the
(α) past error
4/8/2013 Eng. Magdy Abdelsattar 71
72. Exponential Smoothing:
Exponential smoothing is a sales forecasting technique that compares a previous
forecast to actual results to get an error figure to use in current and future forecasts
Trend:
A trend is the upward or downward movement of the numbers in the baseline over
time. Trends indicate some action is necessary, such as ensuring enough inventory is
ordered and enough shipping staff is on hand for high sales months, or additional
sales and marketing efforts are needed for lower sales months. Trends are important
forecasting tools for planning and preparation.
Excel:
Excel is an accounting spreadsheet program that enables users to organize sales
history data for forecasting.
Excel has many features important to sales forecasting, such as pivot tables, averaging
tools and graphing
4/8/2013 Eng. Magdy Abdelsattar 72
73. Why use exponential smoothing?
1. Uses less storage space for data
2. Extremely accurate
3. Easy to understand
4. Little calculation complexity
5. There are simple accuracy tests
Exponential smoothing: the method
Assume that we are currently in period t. We calculated the forecast for
the last period (Ft-1) and we know the actual demand last period (At-1) …
Ft Ft1 ( At1 Ft1 )
The smoothing constant α expresses how much our forecast will react to observed
differences…
If α is low: there is little reaction to differences.
If α is high: there is a lot of reaction to differences.
4/8/2013 Eng. Magdy Abdelsattar 73
74. Where:
Ft = forecast of the time series for the period t
Ft-1 = forecast of the time series for the period t-1
At-1 = actual value of the time series for the period t-1
Ft Ft1 ( At1 Ft1 )
α = the smoothing constant (0 ≤ α ≤ 1)
Or:
F = forecast of the time series for the period t
Ft = forecast of the time series for the period t-1
At = actual value of the time series for the period t-1
F Ft ( At Ft )
α = the smoothing constant (0 ≤ α ≤ 1)
the forecast error in previous period
4/8/2013 Eng. Magdy Abdelsattar 74
75. Forecast Accuracy
An important consideration in selecting a forecasting method is the accuracy of
the forecast.
The mean square error (MSE) is an often-used measure of the accuracy of a
forecasting method.
Week Time Series value Forecast F-Error Square-FE
(t) (Yt) (Ft) (Yt-Ft) (Yt-Ft)^
1 17
2 21 17 4 16
3 19 17.8 1.2 1.44
4 23 18.04 4.96 24.6
5 18 19.03 -1.03 1.06
6 16 18.83 -2.83 8.01
7 20 18.26 1.74 3.03
8 18 18.61 0.61 0.37
MSE= 54.51/8= 6.81……by using α = 0.2
MSE=54.96/8= 6.87…….by using α = 0.3
By trial-and-error we calculate (MSE), The less MSE the most probably forecast we have
4/8/2013 Eng. Magdy Abdelsattar 75
76. Forecast Accuracy
• Mean Absolute Deviation (MAD): measures the total error in a forecast
without regard to sign, Simply the average of the absolute values of all forecast
errors. actual forecast
M AD
n
• Cumulative Forecast Error (CFE): Measures any bias in the forecast.
CFE actual forecast
• Mean Square Error (MSE): Penalizes larger errors.
actual - forecast
2
MS E
n
• Tracking Signal: Measures if your model is working.
CFE
TS
MAD
• Spreadsheet packages are an effective aid is choosing a good value of α for
exponential smoothing and selecting weights for the weighted moving
averages method.
4/8/2013 Eng. Magdy Abdelsattar 76
77. Forecasting Software
1. Spreadsheets
oMicrosoft Excel, Quattro Pro, Lotus 1-2-3
oLimited statistical analysis of forecast data
2. Statistical packages
oSPSS, SAS, NCSS, Minitab
oForecasting plus statistical and graphics
3. Specialty forecasting packages
oForecast Master, Forecast Pro, Autobox, SCA
4/8/2013 Eng. Magdy Abdelsattar 77
78. Which Forecasting Method Should You Use ?
oGather the historical data of what you want to forecast
oDivide data into initiation set and evaluation set
oUse the first set to develop the models
oUse the second set to evaluate
oCompare the MADs and MFEs of each model
4/8/2013 Eng. Magdy Abdelsattar 78
79. Using Excel
1. Using Excel Function
2. Using charting forecasting
3. Using Control Chart
4. Using forecast accuracy
4/8/2013 Eng. Magdy Abdelsattar 79
81. Casual Section
Regression analysis
It is a statistical forecasting method
Process:
consists of identifying causal relationship between company sales (dependent
variable, y) and independent variable (x), which influences sales
If one independent variable is used, it is called linear (or simple) regression, using
formula; y=a + b x, where ‘a’ is the intercept and ‘b’ is the slope of the trend line
In practice, company sales are influenced by several independent variables, like
price, population, promotional expenditure. The method used is multiple regression
analysis
Advantages:
Objective, good accuracy, predicts upturn / downturn, short to medium time, low to
medium cost
Disadvantages:
technically complex, large historical data needed, software packages essential
4/8/2013 Eng. Magdy Abdelsattar 81
82. Regression analysis (terminology)
A statistical technique that can be sued to develop a mathematical equation
showing how variables are related.
• Dependant or response variable: the variable that is being predicted.
• Independent or predictor variables: the variable or variables being used to
predict the value of the dependant variable.
• Simple liner regression: analysis involving one independent variable and
one dependant variable for which the relationship between the variables
is approximated by a straight lin.
• Multiple regression analysis: analysis involving two or more independent
variables.
4/8/2013 Eng. Magdy Abdelsattar 82
83. Liner Regression
• Identify dependent (y) and
independent (x) variables
• Slope of the line
b
XY n X Y
X nX
2 2
• The y intercept
a Y bX
• Develop your equation for the
trend line
Y=(a) + (b) X
4/8/2013 Eng. Magdy Abdelsattar 83
84. Liner Regression
Rest. Y X YX X^
• The slop (b) = 60 1 58 2 116 4
2 105 6 630 36
• The interception (a) = 5 3 88 8 704 64
4 118 8 944 64
• The relation (Y=b+aX) = 5 117 12 1404 144
6 137 16 2192 256
Y = 60 + 5 (X) 7 157 20 3140 400
8 169 20 3330 400
Each time we need to estimate 9 149 22 3278 484
the quarterly sales (Y) knowing 10 202 26 5252 676
the location population we use
Total 1300 140 21040 2528
the above equation.
4/8/2013 Eng. Magdy Abdelsattar 84
85. Correlation Coefficient
• Correlation coefficient (r) measures the direction and strength of
the linear relationship between two variables. The closer the r value
is to 1.0 the better the regression line fits the data points.
n XY X Y
r
X X Y Y
2 2
2 2
n * n
• Coefficient of determination ( r 2 ) measures the amount of variation
in the dependent variable about its mean that is explained by the
regression line. Values of (r 2 ) close to 1.0 are desirable.
4/8/2013 Eng. Magdy Abdelsattar 85
88. Using trend projection in forecasting
• The type of time series for which the trend projection method is applicable shows a
consistent increase or decrease over time.
Y=(a) + (b) X
Y: trend value for sales
(a) : the intercept of the trend line a Y bX
(b) : the slope of the trend line b
XY n X Y
X nX 2 2
4/8/2013 Eng. Magdy Abdelsattar 88
89. Using trend and seasonal components in
forecasting
• How to forecast the values of a time series that has both
trend and seasonal component.
o First step is to compute seasonal indexes SI.
o De-seasonalized the time series by using the SI
o Using regression analysis on the DTS to estimate the trend.
4/8/2013 Eng. Magdy Abdelsattar 89
90. Calculating the seasonal indexes (SI)
year Q1 Q2 Q3 Q4 YA
2003 72 64 63 75 68.5
2004 75 66 64 89 73.5 • First calculate the yearly
average. (YA)
2005 76 68 67 95 76.5
• Calculate the yearly proportions
Yearly proportions
• calculate the overall seasonal
2003 1.051 0.934 0.920 1.095
index for all quarters
2004 1.020 0.898 0.871 1.211
• The seasonal indexes will always
2005 0.993 0.888 0.876 1.242
be the ad up to the number of
time period.
SI 1.021 0.907 0.889 1.183 4
4/8/2013 Eng. Magdy Abdelsattar 90
91. De-seasonalized the time series
year Q1 Q2 Q3 Q4
2003 71 71 71 63
• De-seasonalized is
dividing the actual value
2004 73 73 72 75 by the SI.
2005 74 75 75 80
4/8/2013 Eng. Magdy Abdelsattar 91
93. Estimate the trend
Q value
Using regression analysis on the DTS
X Y YX X^
• The slop (b) = 0.7797
1 71 71 1
2 71 142 4 b
XY n X Y
X nX
2 2
3 71 213 9
4 63 252 16 • The interception (a) = 67.681
5 73 365 25
a Y bX
6 73 438 36
7 72 504 49
• The relation (Y=b+aX) =
8 75 600 64
9 74 666 81
Y = 67.681 + 0.7797 (X)
10 75 750 100
11 75 825 121
12 80 960 144 The value of the time quarter 13
78 873 5786 650 Total (Q1_Y2006) = 77.81
6.50 72.75 482.17 54.17 Average
4/8/2013 Eng. Magdy Abdelsattar 93
94. Seasonal adjustments
• The final step in developing the forecast when
both trend and seasonal components are
present is to use the (SI) of the first Quarter to
adjust the trend projected,
The value of the time quarter 13 (Q1_Y2006) =
77.81 * 1.021 = 79.45
4/8/2013 Eng. Magdy Abdelsattar 94
95. Day tow Recap
oScope of Forecasting
oForecasting Time Horizon
oTypes of Forecasting
4/8/2013 Eng. Magdy Abdelsattar 95
96. Scope of forecasting
Forecasting can be at international level depending upon the
area of operation of particular institution or It can also be
confined to a given product or service supplied by a small firm .
It can be determined in three dimension :
TIME
PRODUCT
GEOGRAPHY
4/8/2013 Eng. Magdy Abdelsattar 96
97. Time Horizon
Short term Forecasts – ( 1-3 Months ) -
These forecasts are tactical decisions. How much inventory should be planned
for next month , how much raw materials to be scheduled for next month
Mid Term Forecasts ( 12-18 months ) -
These are annual plans . How much product should we plan next year? How
much capacities needs to be increased next year ?
Long Term Forecasts ( 5 – 10 Years ) -
These are purely strategic decisions.
What new products to be planned , What new Technology . E.g maruti
planning for mid segment car ( compete with nano)
4/8/2013 Eng. Magdy Abdelsattar 97
98. Types of Forecasting
1) Qualitative –:
These rely on experts opinion in making a prediction for the future .
These are useful for intermediate to Long range forecasting:
o Consumers Survey Methods
o Sales Force Opinion Method
o Delphi Technique
o Scenario Writing
4/8/2013 Eng. Magdy Abdelsattar 98
99. Types of Forecasting
1) Quantitative –
o Time Series - Simple average method, Moving average
o Exponential smoothing
o Linear Regression
o Trend & Seasonal
4/8/2013 Eng. Magdy Abdelsattar 99