Chapter 16
Times Series Forecasting and Index Numbers
McGrawHill/Irwin
Copyright © 2014 by The McGrawHill Companies, Inc. All rights reserved.
Time Series Forecasting
16.1 Time Series Components and Models 16.2 Time Series Regression 16.3 Multiplicative Decomposition 16.4 Simple Exponential Smoothing 16.5 HoltWinter’s Models 16.6 The Box Jenkins Methodology (Optional
Advanced Section)
16.7 Forecast Error Comparisons 16.8 Index Numbers
162
LO16-1: Identify the components of a times series.
16.1 Time Series Components and Models
Trend Cycle
Longrun growth or decline Longrun up and down fluctuation around the trend level Seasonal Regular periodic up and down
movements that repeat within the calendar year
Irregular Erratic very shortrun movements
that follow no regular pattern
163
LO16-1
Time Series Exhibiting Trend, Seasonal, and Cyclical Components
164
Figure 16.1
LO16-1
Seasonality
Some products have demand that varies a
great deal by period ◦Coats, bathing suits, bicycles
This periodic variation is called seasonality ◦Constant seasonality: the magnitude of the swing does not depend on the level of the time series ◦Increasing seasonality: the magnitude of the swing increases as the level of the time series increases
Seasonality alters the linear relationship
between time and demand
165
LO16-2: Use time series regression to forecast time series having linear, quadratic, and certain types of seasonal patterns.
16.2 Time Series Regression
Within regression, seasonality can be modeled
using dummy variables
Consider the model:
yt = (cid:0)
0 + (cid:0)
1t + (cid:0)
Q2Q2 + (cid:0)
Q3Q3 + (cid:0)
Q4Q4 + (cid:0)
t
◦For Quarter 1, Q2 = 0, Q3 = 0 and Q4 = 0 ◦For Quarter 2, Q2 = 1, Q3 = 0 and Q4 = 0 ◦For Quarter 3, Q2 = 0, Q3 = 1 and Q4 = 0 ◦For Quarter 4, Q2 = 0, Q3 = 0 and Q4 = 1
The (cid:0)
coefficient will then give us the seasonal
impact of that quarter relative to Quarter 1 ◦Negative means lower sales, positive lower sales
166
LO16-3: Use data transformations to forecast time series having increasing seasonal variation.
Transformations
Sometimes, transforming the sales data
makes it easier to forecast ◦Square root ◦Quartic roots ◦Natural logarithms
While these transformations can make the forecasting easier, they make it harder to understand the resulting model
167
LO 4: Use multiplicative decomposition and moving averages to forecast time series having increasing seasonal variation.
16.3 Multiplicative Decomposition
We can use the multiplicative decomposition method to decompose a time series into its components:
Trend Seasonal Cyclical Irregular
168
LO 16-5: Use simple exponential smoothing to forecast a time series.
16.4 Simple Exponential Smoothing
Earlier, we saw that when there is no trend, the least squares point estimate b0 of β0 is just the average y value ◦yt = β0 + (cid:0)
t
That gave us a horizontal line that crosses the y axis
at its average value
Since we estimate b0 using regression, each period
is weighted the same
If β0 is slowly changing over time, we want to
weight more recent periods heavier Exponential smoothing does just this
169
LO16-6: Use double exponential smoothing to forecast a time series.
16.5 Holt–Winters’ Models
Simple exponential smoothing cannot handle trend
or seasonality
Holt–Winters’ double exponential smoothing can
handle trended data of the form
t
yt = β0 + β1t + (cid:0) ◦Assumes β0 and β1 changing slowly over time ◦We first find initial estimates of β0 and β1 ◦Then use updating equations to track changes over time Requires smoothing constants called alpha and gamma
1610
LO16-7: Use multiplicative Winters’ method to forecast a time series.
Multiplicative Winters’ Method
Double exponential smoothing cannot handle
seasonality
Multiplicative Winters’ method can handle trended
data of the form yt = (β0 + β1t) ∙ SNt + (cid:0)
t
◦Assumes β0, β1, and SNt changing slowly over time ◦We first find initial estimates of β0 and β1 and seasonal
factors
◦Then use updating equations to track over time
Requires smoothing constants alpha, gamma and delta
1611
LO16-8: Use the Box–Jenkins methodology to forecast a time series.
16.6 The Box–Jenkins Methodology (Optional Advanced Section) Uses a quite different approach Begins by determining if the time series is
stationary ◦The statistical properties of the time series are
constant through time
Plots can help If nonstationary, will transform series
1612
LO16-9: Compare time series models by using forecast errors.
16.7 Forecast Error Comparison
Forecast errors: et = yt ŷt Error comparison criteria
◦Mean absolute deviation (MAD)
n n
y
ˆ y
e t
MAD
(cid:0) (cid:0) (cid:0) t t (cid:0) (cid:0) t t 1 (cid:0) (cid:0)
n
◦Mean squared deviation (MSD)
1 n
y
(
)ˆ y t
MSD
n n 2 (cid:0) (cid:0) (cid:0) 2 e t t (cid:0) (cid:0) t t 1 (cid:0) (cid:0)
n
1613
1 n
LO16-10: Use index numbers to compare economic data over time.
16.8 Index Numbers
Index numbers allow us to compare changes
in time series over time
We begin by selecting a base period Every period is converted to an index by
dividing its value by the base period and then multiplying the results by 100 ◦Simple (quantity) index
t
(cid:0) (cid:0)
100
i t
y y
0
1614

