Chapter 13
Analyzing and
Forecasting Time
Series Data
Chapter 13 - Chapter Outcomes
After studying the material in this chapter, you
should be able to:
•Apply the basic steps in developing and
implementing forecasting models.
•Identify the components present in a time series.
•Use smoothing-based forecasting models
including, single and double exponential
smoothing.
•Apply trend-based forecasting models, including
linear trend, nonlinear trend, and seasonally
adjusted trend.
Forecasting
Model specification refers to the
process of selecting the forecasting
technique to be used in a particular
situation.
Forecasting
Model fitting refers to the process of
determining how well a specified
model fits its past data.
Forecasting
Model diagnosis refers to the process
of determining how well the model
fits the past data and how well the
model’s assumptions appear to be
satisfied.
Forecasting
The forecasting horizon refers to the
number of future periods covered by
the forecast, sometimes referred to as
forecast lead time.
Forecasting
The forecasting period refers to the
unit of time for which the forecasts
are to be made.
Forecasting
The forecasting interval refers to the
frequency with which the new
forecasts are prepared.
Forecasting
Time-Series data are data which are
measured over time. In most applications
the period between measurements is
uniform.
Components of Time Series
Data
• Trend Component
• Seasonal Component
• Cyclical Component
• Random Component
Time Series Forecasting
A time-series plot is a two-dimensional
plot of the time series. The vertical
axis measures the variable of interest
and the horizontal axis corresponds to
the time periods.
Time-Series Plot
(Figure 13-1)
1000
900
800
700
600
$ x 1,000
500
400
300
200
100
0
Time Series Forecasting
A linear trend is any long-term
increase or decrease in a time
series in which the rate of
change is relatively constant.
Time Series Forecasting
A seasonal component is a pattern
that is repeated throughout a time
series and has a recurrence period
of at most one year.
Time Series Forecasting
A cyclical component is a pattern
within the time series that repeats
itself throughout the time series and
has a recurrence period of more than
one year.
Time Series Forecasting
The random component refers to
changes in the time-series data that are
unpredictable and cannot be associated
with the trend, seasonal, or cyclical
components.
Trend-Based Forecasting
Techniques
LINEAR TREND MODEL
yt 0 1t t
where:
yi = Value of trend at time t
0 = Intercept of the trend line
1 = Slope of the trend line
t = Time (t = 1, 2, . . . )
Linear Trend Model
(Example 13-2)
Taft Ice Cream Sales
Year t Sales
1991 1 $300,000
1992 2 $295,000
1993 3 $330,000
1994 4 $345,000
1995 5 $320,000
1996 6 $370,000
1997 7 $380,000
1998 8 $400,000
1999 9 $385,000
2000 10 $430,000
Linear Trend Model
(Example 13-2)
$500,000
Taft Sales
$450,000
$400,000
$350,000
$300,000
Sales
$250,000
$200,000
$150,000
$100,000
$50,000
$0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Year
Linear Trend Model
(Example 13-2)
LEAST SQUARES EQUATIONS
ty n
t
t y t
b1
t ty2
2
b0
y t
b1
t
where: n n
n = Number of periods in the time series
t = Time period independent variable
yt = Dependent variable at time t
Linear Trend Model
(Example 13-2)
SUMMARY OUTPUT
Regression Statistics
Multiple R 0.955138103
R Square 0.912288796
Adjusted R Square 0.901324895
Standard Error 14513.57776
Observations 10
ANOVA
df SS MS F Significance F
Regression 1 17527348485 17527348485 83.20841575 1.67847E-05
Residual 8 1685151515 210643939.4
Total 9 19212500000
Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 277333.3333 9914.661116 27.97204363 2.88084E-09 254470.069 300196.5977 254470.069 300196.5977
t 14575.75758 1597.892322 9.121864708 1.67847E-05 10891.00889 18260.50626 10891.00889 18260.50626
Linear Trend Model
(Example 13-2)
Taft Linear Trend Model
$500,000
$450,000
$400,000
$350,000
y = 14576t + 277333
$300,000
Sales
$250,000
$200,000
$150,000
$100,000
$50,000
$0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Year
Linear Trend Model
- Forecasting -
Trend Projection:
Ft 277,333.33 14,575.76(t )
Forecasting Period t = 11:
Ft 277,333.33 14,575.76(11)
$437,666.69
Linear Trend Model
- Forecasting -
MEAN SQUARE ERROR
MSE
(y t Ft ) 2
n
where:
yt = Actual value at time t
Ft = Predicted value at time t
n = Number of time periods
Linear Trend Model
- Forecasting -
MEAN ABSOLUTE DEVIATION
MAD
|y t Ft |
n
where:
yt = Actual value at time t
Ft = Predicted value at time t
n = Number of time periods
Linear Trend Model
- Forecasting -
MEAN ABSOLUTE DEVIATION
Forecast Bias
(y t Ft )
n
or:
Forecast Bias
(error)
n
Nonlinear Trend Models
(Example)
yt 0 1t t
2
Trend-Based Forecasting
A seasonal index is a number used
to quantify the effect of seasonality
for a given time period.
Trend-Based Forecasting
MUTIPLICATIVE TIME SERIES MODELS
yt Tt St Ct I t
where:
yt = Value of the time series at time t
Tt = Trend value at time t
St = Seasonal value at time t
Ct = Cyclical value at time t
It = Residual or random value at time t
Trend-Based Forecasting
A moving average is the average
of n consecutive values in a time
series.
Trend-Based Forecasting
RATIO-TO-MOVING-AVERAGE
yt
St I t
Tt Ct
Trend-Based Forecasting
DESEASONALIZATION
yt
Tt Ct I t
St
Trend-Based Forecasting
A seasonally unadjusted forecast is a
forecast made for seasonal data that
does not include an adjustment for
the seasonal component in the time
series.
Steps in the Seasonal
Adjustment Process
• Compute each moving average from the k
appropriate consecutive data values.
• Compute the centered moving averages.
• Isolate the seasonal component by
computing the ratio-to-moving-average
values.
• Compute the seasonal indexes by
averaging the ratio-to-moving-averages
for comparable periods.
Steps in the Seasonal
Adjustment Process
(continued)
• Normalize the seasonal indexes.
• Deseasonalize the time series.
• Use least-squares regression to develop the
trend line using the deseasonalized data.
• Develop the unadjusted forecasts using trend
projection.
• Seasonally adjust the forecasts by
multiplying the unadjusted forecasts by the
appropriate seasonal index.
Forecasting Using Smoothing
Techniques
Exponential smoothing is a time-series
smoothing and forecasting technique that
produces an exponentially weighted
moving average in which each smoothing
calculation or forecast is dependent upon
all previously observed values.
Forecasting Using Smoothing
Techniques
EXPONENTIAL SMOOTHING MODEL
Ft 1 Ft ( yt Ft )
or::
Ft 1 yt (1 ) Ft
where:
Ft+1= Forecast value for period t + 1
yt = Actual value for period t
Ft = Forecast value for period t
= Alpha (smoothing constant)
Forecasting Using Smoothing Techniques
DOUBLE EXPONENTIAL SMOOTHING MODEL
Ct yt (1 )(Ct 1 Tt 1 )
Tt (Ct Ct 1 ) (1 )Tt 1
Ft 1 Ct Tt
where:
yt = Actual value in time t
= Constant-process smoothing constant
= Trend-smoothing constant
Ct = Smoothed constant-process value for period t
Tt = Smoothed trend value for period t forecast
value for period t
Ft+1= Forecast value for period t + 1
t = Current time period
Key Terms
• Alpha () • Forecast Error
• Beta () • Forecasting
• Cyclical Component • Forecasting Horizon
• • Forecasting Interval
Deseasonalizing
• Forecasting Period
• Double Exponential
• Linear Trend
Smoothing
• Mean Absolute
• Exponential Deviation (MAD)
Smoothing
• Mean Squared Error
• Forecast Bias (MSE)
Key Terms
(continued)
• Model Diagnosis • Ratio-To-Moving-
• Model Fitting Average Method
• Model Specification • Residual
• Moving Average • Seasonal Component
• Nonlinear Trend • Seasonal Index
• Qualitative Forecasting • Seasonally
• Quantitative Unadjusted Forecast
Forecasting • Smoothing Constant
• Random Component • Splitting Samples
Key Terms
(continued)
• Time-Series Data
• Time-Series Plot
• Trend
