Exponential
smoothing: Pegels
classification
Arum Handini Primandari
Pegels classification
An important consideration in dealing with
exponential smoothing methods having separate
trend and seasonal aspects is whether or not the
model should be additive (linear) or multiplicative
(non linear)
Pegels (1969) has provide a simple but useful
framework for discussing these matters.
Pegels classification: pattern
Pegels classification
Pegels two-way classification is as follows:
Seasonal Component
Trend
component 1 2 3
(none) (additive) (multiplicative)
A
A-1 A-2 A-3
(none)
B
B-1 B-2 B-3
(additive)
C
C-1 C-2 C-3
(multiplicative)
Pegels classification: Formula
All the nine exponential smoothing models can be
summarize by formula:
L t Pt 1 Q t
b t R t 1 b t 1
St Tt 1 St s
where P, Q, R, and T vary according to which of the cells
the method belong to.
Note that:
Cell A-1: SES
Cell B-1: Holts method
Cell B-2: additive Holt-Winters
Cell B-3: multiplicative Holt-Winters
Seasonal Component
Trend 1 2 3
(none) (additive) (multiplicative)
Pt Yt Pt Yt St s Pt Yt / St s
A Q t L t 1 Q t L t 1 Q t L t 1
(none) Ft m L t Tt Yt L t Tt Yt / L t
Ft m L t St m s Ft m L t St ms
Pt Yt Pt Yt St s Pt Yt / St s
Q t L t 1 b t 1 Q t L t 1 b t 1 Q t L t 1 b t 1
B R t L t L t 1
(additive) R t L t L t 1 R t L t L t 1
Ft m L t mb t Tt Yt L t Tt Yt / L t
Ft m L t mb t St m s F L mb S
t m t t t m s
Pt Yt Pt Yt St s Pt Yt / St s
Q t L t 1b t 1 Q t L t 1b t 1 Q t L t 1b t 1
C
R t L t / L t 1 R t L t / L t 1 R t L t / L t 1
(multiplicative)
Ft m L t b mt Tt Yt L t Tt Yt / L t
Ft m L t b mt St m s Ft m L t b mt St m s
Referensi
Makridakris, Wheelwright, dan McGee, 1999,
Metode dan Aplikasi Peramalan Edisi Kedua,
Bumi Aksara.