Forecasting

Principles of Forecasting

Many types of forecasting models that differ in complexity and amount of data & way
they generate forecasts:
1. Forecasts are rarely perfect
2. Forecasts are more accurate for grouped data than for individual items
3. Forecast are more accurate for shorter than longer time periods

Steps in Forecasting Methods

Decide what needs to be forecast

◦ Level of detail, units of analysis & time horizon required
Evaluate and analyze appropriate data
◦ Identify needed data & whether it’s available
Select and test the forecasting model
◦ Cost, ease of use & accuracy
Generate the forecast
Monitor forecast accuracy over time

FORECASTING METHODS

QUANTITATIVE QUALITATIVE OTHER

METHODS METHODS METHODS

TIME SERIES CAUSAL DELPHI FOCUS

METHODS METHODS METHOD FORECASTING

MOVING REGRESSION EXECUTIVE COMBINING

AVERAGES ANALYSIS OPINION METHODS

EXPONENTIAL MULTIPLE MARKET CPFR

SMOOTHING REGRESSION RESEARCH

TREND
PROJECTIONS
Types of Forecasting Methods
Forecasting methods are classified into two groups:

Types of Forecasting Models

Qualitative methods – judgmental methods
◦ Forecasts generated subjectively by the forecaster
◦ Educated guesses
Quantitative methods – based on mathematical modeling:
◦ Forecasts generated through mathematical modeling

Qualitative Methods
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

Time Series Patterns

Time Series Models

Naive:
◦ The forecast is equal to the actual value observed during the last period
◦ good for level patterns
Simple Mean:
◦ The average of all available data
 ◦ good for level patterns
Moving Average:
◦ 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
Weighted Moving Average:
All weights must add to 100% or 1.00
e.g. Ct .5, Ct-1 .3, Ct-2 .2 (weights add to 1.0)
Allows emphasizing one period over others; above indicates more weight on
recent data (Ct=.5)
Differs from the simple moving average that weighs all periods equally - more
responsive to trends
Exponential Smoothing:
Most frequently used time series method because of ease of use and minimal
amount of data needed
Select value of smoothing coefficient, α ,between 0 and 1.0
If no last period forecast is available, average the last few periods or use naive
method
Higher α values (e.g. .7 or .8) may place too much weight on last period’s
random variation

Time Series Problem

Determine forecast for periods 7 & 8
2-period moving average
4-period moving average
2-period weighted moving average with t-1 weighted 0.6 and t-2 weighted 0.4
Exponential smoothing with alpha=0.2 and the period 6 forecast being 375
Period Actual
1 300
2 315
3 290
4 345
5 320
6 360
7 375
8
Time Series Problem Solution
2-
2- 4- Per.Wgted Expon.
Period Actual Period Period . Smooth.
1 300        
2 315       312.0 
3 290  325.5   309.0 295.0
4 345  302.5   300.0 334.0
5 320  317.5 312.5  323.0 325.0
6 360  332.5 317.5 330.0 352.0
7 375 340.0 328.8 344.0 372.0
8   367.5 350.0 369.0 372.6

Forecasting Trend
Basic forecasting models for trends compensate for the lagging that would
otherwise occur
One model, trend-adjusted exponential smoothing uses a three step process

◦ Step 1 - Smoothing the level of the series

S t =αA t +(1−α )( S t−1 +T t−1 )
◦ Step 2 – Smoothing the trend
T t =β( S t −S t−1 )+(1−β )T t−1
◦ Forecast including the trend
FIT t+1=S t +T t

Forecasting trend problem: a company uses exponential smoothing with trend to

forecast usage of its lawn care products. At the end of July, the company wishes to
forecast sales for August. July demand was 62. The trend through June has been 15
additional gallons of product sold per month. Average sales have been 57 gallons per
month. The company uses alpha+0.2 and beta +0.10. Forecast for August.
In addition to α , which is used to smooth out the level of the series, a second coefficient
, β, which is used to smooth out the trend of the series.
Smooth the level of the series:
S July =αA t +(1−α )( St−1 +T t−1 )=( 0 .2 ) ( 62 ) + ( 0 . 8 )( 57+15 )=70
Smooth the trend:
T July=β ( St −S t−1 )+(1−β )T t−1= ( 0. 1 ) (70−57 ) + ( 0 . 9 ) ( 15 )=14 . 8
Forecast including trend:
FIT August =S t +T t =70+14 .8=84 . 8 gallons

Linear Trend Line

A time series technique that computes a forecast with trend by drawing a straight line
through a set of data using this formula:
Y = a + bx
where
Y = forecast for period X
X = the number of time periods from X = 0
A = value of y at X = 0 (Y intercept)
B = slope of the line

Forecasting Seasonality
Calculate the average demand per season
◦ E.g.: average quarterly demand
Calculate a seasonal index for each season of each year:
◦ Divide the actual demand of each season by the average demand per
season for that year
Average the indexes by season
◦ E.g.: take the average of all Spring indexes, then of all Summer
indexes, ...
Forecast demand for the next year & divide by the number of seasons
◦ Use regular forecasting method & divide by four for average quarterly
demand
Multiply next year’s average seasonal demand by each average seasonal index
◦ Result is a forecast of demand for each season of next year
Seasonality problem: a university must develop forecasts for the next year’s quarterly
enrollments. It has collected quarterly enrollments for the past two years. It has also
forecast total enrollment for next year to be 90,000 students. What is the forecast for
each quarter of next year?
Quarter Year 1 Seasonal Year 2 Seasonal Avg. Year3
Index Index Index
Fall 24000 1.2 26000 1.238 1.22 27450
Winter 23000 1.15 22000 1.048 1.10 24750
Spring 19000 0.95 19000 0.905 0.92 20700
Summer 14000 0.70 17000 0.810 0.76 17100
Total 80000 84000 90000
Average 20000 21000 22500

Causal Models
Often, leading indicators can help to predict changes in future demand e.g.
housing starts
Causal models establish a cause-and-effect relationship between independent
and dependent variables
A common tool of causal modeling is linear regression:
Y =a+ bx
Additional related variables may require multiple regression modeling

Linear Regression
 Identify dependent (y) and independent (x) variables
Solve for the slope of the line

b=
∑ XY−n X Y
∑ X 2−n X 2
Solve for the y intercept
a=Y −b X
Develop your equation for the trend line
Y=a + bX

Linear Regression Problem: A maker of golf shirts has been tracking the relationship
between sales and advertising dollars. Use linear regression to find out what sales
might be if the company invested $53,000 in advertising next year.

Sales $ (Y) Adv.$ XY X^2 Y^2

(X)

1 130 32 4160 2304 16,900

2 151 52 7852 2704 22,801

3 150 50 7500 2500 22,500

4 158 55 8690 3025 24964

5 153.85 53

Tot 589 189 28202 9253 87165

Avg 147.25 47.25

b=
∑ XY−n X Y
∑ X 2−n X 2
 28202−4 ( 47 .25 ) (147.25 )
b= 2
=1.15
9253−4 ( 47.25 )
a=Y −b X=147 .25−1 .15 ( 47 .25 )
a=92.9
Y =a+bX=92 .9+1.15X
Y =92.9+1 .15 ( 53 )=153 .85

Correlation Coefficient
How Good is the Fit?
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=
2 2
√n (∑ X )−(∑ X ) ∗√ n (∑ Y )−( Y )
2 2

4 ( 28,202 )−189 ( 589 )

r= =. 982
√ 4(9253 )−(189)2∗√ 4 ( 87,165 )−( 589 )2
r 2 =( . 982 )2=. 964
2
Coefficient of determination ( r ) measures the amount of variation in the
dependent variable about its mean that is explained by the regression line.
2
Values of ( r ) close to 1.0 are desirable.

Multiple Regression
An extension of linear regression but:
◦ Multiple regression develops a relationship between a dependent variable
and multiple independent variables. The general formula is:

Measuring Forecast Error

Forecasts are never perfect
Need to know how much we should rely on our chosen forecasting method
Measuring forecast error:
Et = A t −Ft
Note that over-forecasts = negative errors and under-forecasts = positive errors

Measuring Forecasting Accuracy

 MAD=
∑ |actual−forecast|
Mean Absolute Deviation (MAD) n
◦ measures the total error in a forecast without regard to sign

Cumulative Forecast Error (CFE)

CFE=∑ ( actual−forecast )
◦ Measures any bias in the forecast

∑ ( actual - forecast )2
MSE=
Mean Square Error (MSE) n
◦ Penalizes larger errors

CFE
TS=
Tracking Signal MAD
◦ Measures if your model is working

Accuracy & Tracking Signal Problem: A company is comparing the accuracy of two
forecasting methods. Forecasts using both methods are shown below along with the
actual values for January through May. The company also uses a tracking signal with
±4 limits to decide when a forecast should be reviewed. Which forecasting method is
best?
Month Actual Method A Method B
sales
F’cast Erro Cum Tracking F’cast Error Cum. Tracking
r . Signal Error Signal
Error
Jan. 30 28 2 2 2 27 3 3 1.4

Feb. 26 25 1 3 3 25 1 4 1.8
March 32 32 0 3 3 29 3 7 3.2

April 29 30 -1 2 2 27 2 9 4.1

May 31 30 1 3 3 29 2 11 5

MAD 1 2.2
MSE 1.4 5.4
Selecting the Right Forecasting Model
1. The amount & type of available data
 Some methods require more data than others
2. Degree of accuracy required
 Increasing accuracy means more data
3. Length of forecast horizon
 Different models for 3 month vs. 10 years
4. Presence of data patterns
 Lagging will occur when a forecasting model meant for a level pattern is
applied with a trend

Forecasting Software
Spreadsheets
◦ Microsoft Excel, Quattro Pro, Lotus 1-2-3
◦ Limited statistical analysis of forecast data
Statistical packages
◦ SPSS, SAS, R, NCSS, Minitab
◦ Forecasting plus statistical and graphics
Specialty forecasting packages
◦ Forecast Master, Forecast Pro, Autobox, SCA

Other Forecasting Methods

Focus Forecasting - Focus forecasting uses a computer simulation program
that evaluates the forecast performance of a number of rules on past data. The
program keeps track of the rules and evaluates how well they perform.
◦ Developed by Bernie Smith
◦ Relies on the use of simple rules
◦ Test rules on past data and evaluate how they perform
Combining Forecasts
◦ Combining two or more forecasting methods can improve accuracy

Collaborative Planning Fore-casting & Replenishment (CPFR)

The premise behind CPFR is that companies can be more successful if they join
forces to bring value to their customers, share risks of the marketplace, and
improve their performances.
◦ Establish collaborative relationships between buyers and sellers
 ◦ Create a joint business plan
◦ Create a sales forecast
◦ Identify exceptions for sales forecast
◦ Resolve/collaborate on exception items
◦ Create order forecast
◦ Identify exceptions for order forecast
◦ Resolve/collaborate on exception items
◦ Generate order

Forecasting within OM: How it all fits together

Forecasts impact not only other business functions but all other operations decisions.
Operations managers make many forecasts, such as the expected demand for a
company’s products. These forecasts are then used to determine:
product designs that are expected to sell,
the quantity of product to produce,
the amount of needed supplies and materials.
Also, a company uses forecasts to
determine future space requirements,
capacity and
location needs, and
the amount of labor needed.
Forecasts drive strategic operations decisions, such as:
choice of competitive priorities, changes in processes, and large technology
purchases.
Forecast decisions serve as the basis for tactical planning; developing worker
schedules.
Virtually all operations management decisions are based on a forecast of the future.
Forecasting Across the Organization
Forecasting is critical to management of all organizational functional areas
◦ Marketing relies on forecasting to predict demand and future sales
◦ Finance forecasts stock prices, financial performance, capital investment
needs.
◦ Information systems provides ability to share databases and information
◦ Human resources forecasts future hiring requirements

Highlights
Three basic principles of forecasting are: forecasts are rarely perfect, are more
accurate for groups than individual items, and are more accurate in the shorter
term than longer time horizons.

The forecasting process involves five steps: decide what to forecast, evaluate
and analyze appropriate data, select and test model, generate forecast, and
monitor accuracy.

Forecasting methods can be classified into two groups: qualitative and

quantitative. Qualitative methods are based on the subjective opinion of the
forecaster and quantitative methods are based on mathematical modeling.
Time series models are based on the assumption that all information needed is
contained in the time series of data. Causal models assume that the variable
being forecast is related to other variables in the environment.

There are four basic patterns of data: level or horizontal, trend, seasonality, and
cycles. In addition, data usually contain random variation. Some forecast models
used to forecast the level of a time series are: naïve, simple mean, simple
moving average, weighted moving average, and exponential smoothing.
Separate models are used to forecast trends and seasonality.

A simple causal model is linear regression in which a straight-line relationship is

modeled between the variable we are forecasting and another variable in the
environment. The correlation is used to measure the strength of the linear
relationship between these two variables.

Three useful measures of forecast error are mean absolute deviation (MAD),
mean square error (MSE) and tracking signal.

There are four factors to consider when selecting a model: amount and type of
data available, degree of accuracy required, length of forecast horizon, and
patterns present in the data.