Demand Forecasting
(Part Two)
Harry Kogetsidis
School of Business
Lecture’s topics
• How can forecasting methods be used to
predict demand?
• How does the method of Holt’s exponential
smoothing work?
• How do we measure forecast accuracy?
Holt’s Exponential Smoothing
An extension of simple exponential smoothing
which provides a more appropriate forecasting
method for time series that have a trend.
trend component
Like simple exponential smoothing, this method
also aims to smooth out random fluctuations
in the data values.
irregular component
Holt’s Exponential Smoothing
deals with random fluctuations
L
t
= Y
t
+ (1-)(L
t-1
+ T
t-1
)
(1)
deals with trend
T
t
= (L
t
– L
t-1
) + (1-)T
t-1
(2)
Holt’s Exponential Smoothing
smoothing constant
L
t
= Y
t
+ (1-)(L
t-1
+ T
t-1
)
(1)
smoothing constant
T
t
= (L
t
– L
t-1
) + (1-)T
t-1
(2)
Holt’s Exponential Smoothing
L
t
= Y
t
+ (1-)(L
t-1
+ T
t-1
)
(1)
T
t
= (L
t
– L
t-1
) + (1-)T
t-1
(2)
F
t+1
= L
t
+ T
t
(3)
Holt’s Exponential Smoothing
L
t
= Y
t
+ (1-)(L
t-1
+ T
t-1
)
(1)
T
t
= (L
t
– L
t-1
) + (1-)T
t-1
(2)
F
t+1
= L
t
+ T
t
(3)
forecast for period t+1
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
Jan 20
20.0
20.0
0.0
Feb 24
20.0
Mar 27 23.2
Apr 31 26.2
May 37 30.1
Jun 47 35.6
Jul
53 44.7
Aug 62 51.3
Sep
59.9
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
Jan 20
20.0
20.0
0.0
Feb 24
20.0
Mar 27 23.2
Apr 31 26.2
May 37 30.1
Jun 47 35.6
Jul
53 44.7
Aug 62 51.3
Sep
59.9
L
t
= Y
t
+ (1-)(L
t-1
+ T
t-1
)
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
Jan 20
20.0
20.0
0.0
Feb 24
20.0
22.0
Mar 27 23.2
Apr 31 26.2
May 37 30.1
Jun 47 35.6
Jul
53 44.7
Aug 62 51.3
Sep
59.9
T
t
= (L
t
– L
t-1
) + (1-)T
t-1
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
Jan 20
20.0
20.0
0.0
Feb 24
20.0
22.0 1.0
Mar 27 23.2
…
…
Apr 31 26.2
…
…
May 37 30.1
…
…
Jun 47 35.6
…
…
Jul
53 44.7
…
…
Aug 62 51.3
…
…
Sep 59.9
…
…
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
Jan 20
20.0
20.0
0.0
-
Feb 24
20.0
22.0 1.0
Mar 27 23.2
25.0
2.0
Apr 31 26.2
29.0
3.0
May 37 30.1
34.5
4.3
Jun 47 35.6
42.9
6.3
Jul
53 44.7
51.1
7.3
Aug 62 51.3
60.2
8.2
Sep 59.9
F
t+1
= L
t
+ T
t
Holt’s exponential smoothing (=0.5,
=0.5)
Period
Y
t
F
t
L
t
T
t
F
t
|e
t
| |e
t
/Y
t
|
Jan 20
20.0
20.0
0.0
-
- -
Feb 24
20.0
22.0 1.0
20.0
Mar 27 23.2
25.0
2.0
23.0
Apr 31 26.2
29.0
3.0
27.0
May 37 30.1
34.5
4.3
32.0
Jun 47 35.6
42.9
6.3
38.8
Jul
53 44.7
51.1
7.3
49.2
Aug 62 51.3
60.2
8.2
58.4
Sep 59.9
68.4
Comparing the two models
MAD
MAPE
• Simple exponential smoothing (=0.8)
• Holt’s exponential smoothing (=0.5, =0.5)
Document Outline
- Demand Forecasting (Part Two)
- Lecture’s topics
- Holt’s Exponential Smoothing
- Holt’s Exponential Smoothing
- Slide 5
- Slide 6
- Slide 7
- Holt’s exponential smoothing (=0.5, =0.5)
- Slide 9
- Slide 10
- Slide 11
- Slide 12
- Slide 13
- Comparing the two models
Wyszukiwarka
Podobne podstrony:
Section 2 student notes
Section 9 student notes
Section 6 student notes
Section 3 student notes
Section 4 student notes
Section 5 student notes
Section 1 student notes
Section 8 student notes
Section 10 student notes
Section 11 student notes
Posadzki section 2 1 studenci
Posadzki section 2 3 studenci
Posadzki section 3 studenci
Posadzki section 1 studenci
Posadzki section 2 studenci
Posadzki section 2 2 studenci
17 Steps to Better Presentations Student Notes
więcej podobnych podstron