Professional Documents
Culture Documents
Overview
Introduction
Smoothing techniques
Trend analysis
Analysis
Forecasting
Application
Example:
Detect patterns
to forecast the
future value of
the time-series
Applications in
management and
economics
Forecast
interest rates,
U/E rate
Predict the
demand for
products
Long-term
trend
Cyclical
effect
(T)
(C)
Seasonal
effect
Random
variation
(S)
(R)
Volume
Time
9
10
Volume
Random
variation that
does not repeat
Time
11
2 types:
Exponential smoothing
12
Example of
Moving
average:
Period
t
yt
3-period
MA
4-period
MA
4-period centred
MA
12
18
15.33
16
19.33
17.5
18.13
24
19.00
18.75
18.50
17
19.00
18.25
19.38
16
19.33
20.5
20.13
25
20.67
19.75
21
13
Linear model:
Techniques
yt = 0 + 1t +
Polinomial
model
Trend analysis
Purpose
Isolate the
long-term trend
14
Quarter
Year
Calculate
2005
1
2
MAt : Mulplicative
model:
-
1.0254
0.9572
1.0281
1.0318
0.9548
1.0316
1.0212
0.9592
1.0134
1.0481
2006
2007
MAt = T0.9869
t x Ct
2008
2009
1.0012
T x0.9304
Ct x St -x Rt
Yt
0.9900t
Average (Si)
0.9925
Seasonal Index
t (Si)
0.9928
MA
4 Total
1.0239
yt = Tt x Ct x St x Rt
0.9918
0.9504
1.0242
T
0.9507
t x Ct1.0246
1.0316
3.9987
1.0319
4.0000
Calculate average of St x Rt
St
St is adjusted SIt , so that
average SIt= 1
where:
Ft = forecast for period t
SIt = seasonal index for period t
16
Year
2005
2006
2008
Month
1
2
3
4
5
6
7
8
9
10
11
12
48
CPI
101.1
102.5
100.1
100.6
100.5
100.4
100.4
100.4
100.8
100.4
100.4
100.8
99.3
17
18
Reasons:
3 components exist
regression
19
CPI peaks in
Feb
Random
variation in 2008
20
Trend analysis
Using Excel, the trend line is:
yt = 100.551 + 0.016 t
y = 100.551 + 0.016 t
21
yt = Tt x Ct x St x Rt
MAt = Tt x Ct
Yt
MAt
Tt x Ct x St x Rt
Tt x Ct
Calculate average of St x Rt
St
St is adjusted SIt , so that
average SIt= 1
22
Seasonal
index
Month
SI t
1.0055
1.017
0.997
Month
SI t
11
12
Apply
23
25
y = 100.551 + 0.016 t
Forecasted
CPI