EM6113-Engineering Management

Techniques

Forecasting

1
Forecasting
 Essential preliminary to effective planning

 Engineering manager must be concerned

with both future markets and future
technology
Why Forecasting?
 New facility planning

 Production planning

 Work force scheduling

Long Range Forecasts
 Design new products

 Determine capacity for new product

 Long range supply of materials

Short Range Forecasts
 Amount of inventory for next month
 Amount of product to produce next week
 How much raw material delivered next week
 Workers schedule next week
Types of Forecasting Methods

 Forecasting methods are classified into two

groups:

© Wiley 2010 6
Forecasting
Qualitative Methods
 Judgment Methods
 Jury of Executive Opinion
 Delphi
 Sales Force Composite
 Users’ Expectation (Surveys)
Jury of Executive Opinion
 Simplest method
 Executives provide an estimate
 Educated guess
 Average of estimates taken
Delphi Method
 Eliminates effects of interactions between
members
 Experts do not need to know who other
experts are
 Delphi coordinator asks for opinions,
forecasts on subject through questionnaires
Delphi Method, cont
 Develop objective of forecast
 Determine number of participants
 Select and contact participants
 Develop first questionnaire and submit
 Coordinator analyzes responses
 Delphi Method, cont
 Develop second questionnaire based on
results of first
 Share aggregate results of first round
 Analyze responses
 This technique eliminates the effects of
interaction among experts
 Rounds continue until consensus reached or
experts’ opinions cease to change
 Sales Force Composite
 Members of the sales force estimate sales
in their own territory
 Regional Sales Managers adjust for their
opinion of the optimism or pessimism of
individual sales people
 General Sales Manager ‘massages’ the
figures to account for new products or
factors others are unaware of
Users’ Expectation (Market
research)
 When small customer base, simplest method
is to ask customers to project their need for
the future period
 Market testing or market surveys
 Information expensive to obtain
 Often customers don’t know their future need
 Qualitative Methods

Type Characteristics Strengths Weaknesses

Executive A group of managers Good for strategic or One person's opinion
opinion meet & come up with new-product can dominate the
a forecast forecasting forecast

Market Uses surveys & Good determinant of It can be difficult to

research interviews to identify customer preferences develop a good
customer preferences questionnaire

Delphi Seeks to develop a Excellent for Time consuming to

method consensus among a forecasting long-term develop
group of experts product demand,
technological
© Wiley 2010 14
changes, and
Forecasting
Quantitative Methods
 Time Series Methods
 Moving Average
 Weighted Moving Average
 Exponential Smoothing

 Association or Causal Method

 Simple Regression
 Multiple Regression
Quantitative Methods
 Time Series Models:
 Assumes information needed to generate a forecast is
contained in a time series of data
 Assumes the future will follow same patterns as the
past
 Causal Models or Associative Models
 Explores cause-and-effect relationships
 Uses leading indicators to predict the future
 Time Series Models
 Forecaster looks for data patterns as
 Data = historic pattern + random variation
 Historic pattern to be forecasted:
 Level (long-term average): data fluctuates around a constant mean
 Trend: data exhibits an increasing or decreasing pattern
 Seasonality: any pattern that regularly repeats itself and is of a
constant length
 Cycle: patterns created by economic fluctuations
 Random Variation cannot be predicted

© Wiley 2010 17
Time Series Patterns

© Wiley 2010 18
 Time Series Models

 Naive: Ft 1  At
 The forecast is equal to the actual value observed during
the last period – good for level patterns
 Simple Mean: Ft 1   A t / n
 The average of all available data - good for level patterns
 Moving Average: Ft 1   A t / n
 The average value over a set time period
(e.g.: the last four weeks)
 Each new forecast drops the oldest data point & adds a
new observation
 More responsive to a trend but still lags behind actual data

© Wiley 2010 19
 Simple Moving Average
 Forecast Ft is average of n previous
observations or actuals At :

 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;
 Simple Moving Average
 Include n most recent observations
 Weight equally
 Ignore older observations
weight

1/n

n ... 3 2 1
today
Time Series Models con’t
 Weighted Moving Average:
Ft 1   Wt A t

 All weights must add to 100% or 1.00

e.g. Wt .5, Wt-1 .3, Wt-2 .2 (weights add to 1.0)

 Allows emphasizing one period over others; above

indicates more weight on recent data (Wt=.5)

 Differs from the simple moving average that weighs all

periods equally - more responsive to trends
 Exponential Smoothing: Math
Fn  1  Fn   ( An  Fn )
Fn  1   An  (1   ) Fn
The forecast value for the next period is Fn 1
taken as the sum of
• The forecasted value for the current period Fn
• Some fraction of the difference between the actual and forecasted
values for the current period
• In order to understand graphical representation:

F n  1   A n   (1   ) A n  1   (1   ) A n  2  
2

F n  1   A n  (1   )  A n  1   (1   ) A n  2   
 Exponential Smoothing:
 Include all past observations
 Weight recent observations much more heavily
than very old observations:
weight 0    1
Decreasing weight given 
to older observations
(1  )
(1  )2
(1  ) 3

today
 Exponential Smoothing:
 Thus, new forecast is weighted sum of old
forecast and actual value
 Notes:
 Only 2 values (An and Fn ) are required, compared
with n for moving average
 Parameter  determined empirically (whatever
works best)
 Rule of thumb:  < 0.5
 Typically,  = 0.2 or  = 0.3 work well
Regression Analysis

Dependent
variable

Independent variable (x)

Regression is the attempt to explain the variation in a dependent
variable using the variation in independent variables.
Regression is thus an explanation of causation.
If the independent variable(s) sufficiently explain the variation in the
dependent variable, the model can be used for prediction.
 Simple Linear Regression

y’ = b0 + b1X ± є
variable (y) є
Dependent

B1 = slope
b0 (y intercept) = ∆y/ ∆x

Independent variable (x)

To recall think of the equation y=c+mx where m is slope
The output of a regression is a function that predicts
the dependent variable based upon values of the
independent variables.
Simple regression fits a straight line to the data.
What’s Slope?

A slope of 2 means that every 1 unit change in

X yields a 2 unit change in Y.
What’s Prediction

If you know something about X, this knowledge

helps you predict something about Y.
 Simple Linear Regression

Observation: y
Prediction: ^y
Dependent
variable

Zero
Independent variable (x)

The function will make a prediction for each observed

data point.
The observation is denoted by y and the prediction is
denoted by ^y.
 Simple Linear Regression

Prediction
error: ε
Observation: y
Prediction: ^y

Zero

For each observation, the variation can be described as:

^y = y + ε

Actual = Explained + Error

 Regression

Dependent
variable

Independent variable (x)

A least squares regression selects the line with the
lowest total sum of squared prediction errors.
This value is called the Sum of Squares of Error, or
SSE.
 Multiple Linear Regression

More than one independent variable can be used to explain

variance in the dependent variable, as long as they are not
linearly related.

A multiple regression takes the form:

y = A + β X + β X + … + β k Xk + ε
1 1 2 2

where k is the number of variables, or parameters.

 Multicollinearity

Multicollinearity is a condition in which at least 2

independent variables are highly linearly correlated. It
will often crash computers.
Example table of
Correlations
Y X1 X2
Y 1.000
X1 0.802 1.000
X2 0.848 0.578 1.000

A correlations table can suggest which independent

variables may be significant. Generally, an ind. variable
that has more than a .3 correlation with the dependent
variable and less than .7 with any other ind. variable can
be included as a possible predictor.
 Nonlinear Regression

Nonlinear functions can also be fit as regressions.

Common choices include Power, Logarithmic,
Exponential, and Logistic, but any continuous
function can be used.
Which Method?
 Select a few methods

 Make forecasts

 Take simple average

 No one best answer

Forecasting New Products
 First use judgmental

 Expert opinions

 Consumer intentions
Technological Forecasting
 Which technologies will be available in the
future
 Part of planning as this is the context in which
plans will operate or be implemented
 Delphi Method
 Internet
 Cookies
 Demographics
 Trends
Strategies for Managing
Technology
 Invention and Innovation
 Entrepreneurship
 Intrapreneurship
 Managing Technological Change
 E.g. advent of the Internet
 Government Regulations
Questions?