Accepted Manuscript

A Comparison Study between Fuzzy Time Series Model and ARIMA Model
for Forecasting Taiwan Export
Chi-Chen Wang
PII:
DOI:
Reference:

S0957-4174(11)00035-2
10.1016/j.eswa.2011.01.015
ESWA 5759

To appear in:

Expert Systems with Applications

Please cite this article as: Wang, C-C., A Comparison Study between Fuzzy Time Series Model and ARIMA Model
for Forecasting Taiwan Export, Expert Systems with Applications (2011), doi: 10.1016/j.eswa.2011.01.015

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

A Comparison Study between Fuzzy Time Series Model and ARIMA Model for
Forecasting Taiwan Export
Chi-Chen Wang*
Department of Financial Management, National Defense University ,No 70,sec.2,Zhongyang N, Rd,
Beitou District Pei-Tou , Taipei, Taiwan, R.O.C.

Abstract
This study compares the application of two forecasting methods on the amount
of Taiwan export, the ARIMA time series method and the fuzzy time series method.
Models discussed for the fuzzy time series method include the Factor models, the
Heuristic models, and the Markov model. When the sample period is prolong in our
models, the ARIMA model shows smaller than predicted error and closer predicted
trajectory to the realistic trend than those of the fuzzy model, resulted in more
accurate forecasts of the export amount in the ARIMA model. Especially, the
coefficient of the error term for the previous period has increased to 79%, implying
the influential effect of external factors. These external factors attribute to the export
amount of Taiwan according to the economic viewpoints. However ,this impact
reduces as time progressing and the export amount of the lag period of 12 or 13 do not
affect current export amount anymore. In conclusion, when the sample period is
shorter with only a small set of data available, the fuzzy time series models can be
utilized to predict export values accurately, outperforming the ARIMA model .

Key words: ARIMA model, fuzzy time series, Taiwan export.

Associate Professor, E-mail: [email protected] ,Tel:886-2-28966301 ext.604989.

I. Introduction
Owing to the relatively small domestic market and the scarce nature resources in
Taiwan, the economic development of Taiwan relies heavily on the international
trades. In the past, Taiwan used its low labor cost advantage to manufacture
labor-intensive products as a competitive edge to expand export worldwide and
gained tremendous economic growth. However, the low labor cost advantage has no
longer existed in recent years that the standard wage is comparatively much higher in
Taiwan than in the mainland China and Southeast Asia. Taiwan is facing with sever
challenges from these lower labor cost countries in the international markets and the
amount of export has shrunken rapidly. Consequently, over the past decade, Taiwans
economic is stagnating and the unemployment rate reaches its highest record of over 6
percent. Given the highly economic dependant nature on the export business of
Taiwan, better understanding toward future trends in the international business to
enhance market competition is an important task. Therefore, it is vital to have more
accurate forecast and prediction models to assist and enact economic policies and
business strategies particularly in the international export trade.
In recent years, due to the innovation of the forecasting technique and
improvement of the forecasting accuracy, the forecasting methodology becomes
indispensable for further decision making process in both industry and government.
However, from the viewpoints of model base, traditionally, most of forecasting
methods usually construct a statistical model as a tool for forecasting future value and
data analysis. Within the forecasting process, it aims to find out a best model through
the historical data of pretest groups. Nevertheless, due to the error caused in data
collection, time lag and reciprocal effect between variables may make the data itself
containing high uncertainty. For instance, in forecasting currency exchange rate
between NT dollars and US dollars, data from opening quotation, including closing
quotation, highest, lowest or average price, may cause the forecasting result different.
These time series data may appear to be exact numbers, but actually they represent
probable values within their intervals. For example, stock price fluctuation in the
market could be communicated in linguistic terms such as asking someones mood,
which may cost the data containing high uncertainty. Consequently, problems of
over-fit the model and over-explained the data may result by using traditional
methods in the analysis of a statistical model.
Zadeh (1965) first defines a fuzzy algorithm is an ordered set of fuzzy
instructions which upon execution yield an approximate solution to a specified
problem. The concept of fuzzy set is wildly applied to different fields and to solve
the problems of linguistic data. Recently, the idea of fuzzy logic has been successfully
applied in dynamic analysis forecasting methods. For instance, based on fuzzy theory,
2

Song and Chissom (1993a, 1993b, 1994) construct the one-factor fuzzy time series.
Sullivan and Woodall (1994) review two related models, the first-order time-invariant
fuzzy time series model and the first-variant model proposed by Song and Chissom,
where the two models are compared with each other and with a time-invariant
Markov model using linguistic labels with probability distributions. Chen (1996)
proposes a new concept of time series method which is more efficient than the one
presented in Song and Chissom (1993a). The proposed method by Chen (1996) uses
simplified arithmetic operations rather than those complicated maximum composition
operations in Song and Chissom (1993a). In Chens method, the variation of
enrollment of this year is related to the trend of past years. To define the degrees of
variations, he performs systematic calculations to obtain the relation between the
variation of last year and other previous years. Then, he can get the forecasting
enrollments from the derived relation. Huarng (2001a, 2001b) proposes Heuristic
models by integrating problem-specific heuristic knowledge with Chens model to
improve forecasting, since Chens model is easy to calculate, straightforward to
integrate heuristic knowledge, and could forecast better than others. Lee et al. (2006),
based on the two-factor high-order fuzzy time series and historical data, proposes
two-factor high-order fuzzy logical relationships to increase the forecasting accuracy
rate. Lee et al.(2008) base on high-order fuzzy logical relationships and genetic
simulated annealing techniques. Their proposed method uses genetic simulated
annealing techniques to adjust the length of each internal in the universe of discourse
to increase the forecasting accuracy rate. However, few studies pay attention to
discuss the forecasting ability of fuzzy time series and traditional time series model.
Besides, previous researches only compute the forecasting data using a certain length
of interval, and neglect the fact that differences in the time periods may probably
influence the outcome as well.
This research attempts to use information of export values in Taiwan as an example
to test whether the fuzzy time series is indeed practical in its forecast of
macro-economic variables. By comparing the fuzzy time series with the ARIMA time
series method, better understanding of the appropriateness of forecasting models
could be obtained. Furthermore, this paper intents to identify which model is able to
provide information in decision making, especially when the data is insufficient. The
results could improve the understandings of how time periods included in data affects
forecasting performance, by analyzing the fuzzy time series and the ARIMA time
series model.
II. RESEARCH METHOD
1.ARIMA model
The model was propose by BoxJenkins in 1970, which examines each variable
3

by using auto-regression, AR (P) and Moving Average, MA (Q) to investigate the

historical data and economic fluctuations. The algorithm presented as follows
(1) Data preparation.
(2) The first step in developing a BoxJenkins model is to determine if the series is
stationary and if there is any significant seasonality that needs to be modeled. The
autocorrelation functions (ACF) are used to define the distribution of sample data.
(3) Differencing to achieve stationary.
(4) Identification: identifying the phase of the series by using autocorrelation
functions (ACF) and partial autocorrelation function (PACF).
(5) Estimation: the conditional likelihood and exact likelihood are used to estimate the
parameters.
(6) Diagnostic check: the process of diagnostic check involves testing the assumptions
of the model to identify any areas where the model is inadequate. The statistical
identification process includes: whether the parameters achieve statistical significance
or multicollinearity and whether the residual term is white noise or not. If the model is
found to be inadequate, it is necessary to remedy and repeat Step( 4) until a better
model is identified.

2. Fuzzy time series models

2.1 Fuzzy time series concepts
Definition2.1.1 A fuzzy set is a class of objects with a continuum of grade of
membership. Let U be the universe of discourse with U = {u1 , u 2 ,....u t } .where u i
are possible linguistic values of U , then a fuzzy set of linguistic variables Ai of U is
defined by
Ai = f Ai (u1 ) / u1 + f Ai (u 2 ) / u 2 + ............ + f Ai (u t ) / u t , where f Ai is a membership
function of fuzzy set, Ai , f Ai : U [0,1] . u k is an element of fuzzy set Ai ,and
f Ai (u k ) is the degree of belongingness of u k to Ai . f Ai (u k ) [0,1] and 1 k n .

Definition 2.1.2 Let Y (t )(t = ....,0,1,2,.....), is a subset of R. Let Y (t ) be the universe

of discourse defined by the fuzzy set f i (t ) . If F (t ) consists of
f i (t )(i = 1,2,......), F (t ) is defined as a fuzzy time series on Y (t )(t = ....,0,1,2,.....).
Definition2.1.3 Suppose F (t ) is caused only by F (t 1) and is denoted by
F (t 1) F (t ) ; then there is a fuzzy relationship between F (t ) and F (t 1)
which can be expressed as the fuzzy relational equation: F (t ) = F (t 1) o R(t , t 1)

where o is Max-Min composition operator. The relation R is called the first-order

of F (t ) . Further if fuzzy relation R (t , t 1) of F (t ) is independent of time t ,
4

meaning for different times t1 and t 2 , R(t1 , t 1) = R(t 2 , t 2 1), then F (t ) is

called a time invariant fuzzy time series.

2.2 Factors time-variant fuzzy time series model

Definition 2.2.1 The criterion matrix C(t) and the operation matrix O w (t) at year t are
expressed as follows:
C (t ) = f (t 1) = [C1 , C 2 , L , C m ] , 0 C j 1

(1)

f (t 2) O11 O12 L O1m

f (t 3) O
O22 L O2m
21
w

0 Oij 1 , 1 i w 1 , 1 j m (2)
O (t ) =
=
M ,
M M

f (t w) O( w1)1 L L O( w1)m

Where f (t 1) is the fuzzified variation of the first factor F (t ) between

times (t 1) and (t 2) ; m the number of intervals in the universe of discourse,
w the window basis.
Definition 2.2.2 Relationship matrix R(t )
R(t ) = O w (t ) C (t )
O11 C1
O C
21
1
=

O( w1)1 C1
R11
R
21
=
M

R( w1)1

O12 C 2
O22 C 2
M
O( w1) 2 C 2

R12
R21

.L
L

M
L

R( w1) 2

O1m C m
O2 m C m

M
M

L O( w1) m C m
L
L

R1m
R2 m
M

R( w1) m

0 Oij 1 , 1 i w 1 , 1 j m , 0 C j 1

(3)

Where Rij = Oij C j , is the multiplication operator. From the matrix model (3),
the fuzzified variation f (t ) can be described as follows:

f (t) = max(R11, R21,...,R(w1)1 ), max(R12 , R22 ,...,R(w1)2 ),...,max(R1m , R2m ,...,R(w1)m )

(4)

The model algorithm for the two-factor fuzzy time series is presented as follows:
Step1. Decide the first factor and calculate its variation.
Step2. Decide universe of discourse and the length of intervals.
5

Step3. Decide the window size for the fuzzy relationship matrix.
Step4. Compute the fuzzified variation and defuzzify.

2.3 Heuristic fuzzy time series model

Definition 2.3.1 Heuristic fuzzy logical relationship
Suppose that there is a relationship between series F (t-1) and F (t), it is defined
as F (t ) = F (t 1) R(t 1, t ) , where is an operator. Further, the fuzzy logical
relationship between series F (t ) and F (t 1) is expressed by

(5)
F (t 1) F (t )
Let F (t) =Aj and F (t-1) =Ai , a fuzzy logical relationship is defined as Ai A j ,
where Ai is named as left-hand side, Aj the right-hand side.

Definition 2.3.2 Heuristic fuzzy logical relationships groups

The fuzzy logical relationship can be further integrated into fuzzy groups
because on the same left-hand sides. For example, there are fuzzy logical relationships
with the same left-hand sides ( Ai ):
Ai A j1 Ai A j 2 Ai A j 3
,
,

(6)

Therefore, these fuzzy logical relationships can be expressed as follows:

Ai Aj1 , Aj 2 , Aj 3

(7)

Definition 2.3.3 In the heuristic models, heuristic function takes fuzzy logical
relationship groups and relevant variables as parameters. From these fuzzy logical
relationship groups, the heuristic functions use the variables to select proper fuzzy
sets to establish heuristic fuzzy logical relationship groups.
All the fuzzy sets A1, A2 , A3 ,L , Ak are well ordered. This condition greatly facilitates
the selection of proper fuzzy sets by the heuristic function.
Suppose F (t 1) = Ai and the fuzzy logical relationship group for Ai is
Ai A j1 , A j 2 , A j 3 L

Proper fuzzy sets, A p1 , A p 2 , A ps , L A pk , can the selected by heuristic function h(x),

h( x; A j1 , A j 2 , A j 3 ,L) = A p1 , A p 2 , A p 3 , L A pk ,

where

are

heuristic

variable; A p1 , A p 2 , A p 3 ,L A pk , are selected from A j1 , A j 2 , A j 3 ,L ,by the heuristic

function. A heuristic fuzzy logical relationship group is obtained below:

The model algorithm of heuristic fuzzy time series can be presented as follows:
step1. Decide the universe of discourse and the length of interval.
step2. Define the fuzzy sets and establish the fuzzy logical relationships.
Step3. Define the second and third variable.
Step4. Redefine the fuzzy logical groups and the threshold auto regression model.
Step5. Forecast and defuzzify.

2.4 Markov fuzzy time series model

Definition 2.4.1 Markov fuzzy relation
Let U be the universe of discourse, with G = ( 1 , L , r )
and {Pi , i = 1,2, L, r} defined as an ordered partition of U . i
membership functions on the universe of the fuzzy set U . The
fuzzy groups G and H is defined as

, H = (v1 , L , v r ) ,
and vi are the
relation between

[ ]

R = G T o H = Rij

(8)

r r

Where o is the max-min operator, T the transpose, Rij the membership function
between G and H.

Definition 2.4.2 Markov fuzzy relation matrix

It is assumed that { F ( X (t )) } is a fuzzy autoregressive process of order one (FAR
(1)), meaning that any one of t, F ( X t ) completely depends on F ( X t 1 ) . Let the

membership function of F ( X t ) be i ( X t ) , i = 1,2,..., r , then the Markov fuzzy

relation matrix is represented by

[ ]

* = *ij

rr

= max min ( i (X t 1 ), j (X t )) r r
2 t n

( 9)

Definition 2.4.3 Fuzzy auto-regressive model of order p

If the fuzzy time series model is expressed as below:
F (t ) = F (t 1) o F (t 2 ) o ... o F (t p ) *

(10)
*
p

then we call it a fuzzy auto-regressive model of order

, where
is the fuzzy
(
)
(
)
(
)
(
)
F
t
F
t

1
F
t

2
F
t

p
.
and
,
,,
relation matrix between

Definition 2.4.4 First-order multi-variant fuzzy auto-regressive series

The series ( FX 1,t , FX 2,t ,..., FX k ,t ) is defined as first-order multi-variant fuzzy
auto-regressive if the equation below is satisfied.

11 L 1k
( FX 1,t , FX 2,t ,..., FX k ,t ) = ( FX 1,t 1 , FX 2,t 1 ,..., FX k ,t 1 ) M O M
k1 L kk

(11)

i = 1,2,..., k j = 1,2,..., k
Where ij is the Markov fuzzy relation matrix. Since the series
( FX 1,t , FX 2,t ,..., FX k ,t )
completely
depends
on
the
status
of
( FX 1,t 1 , FX 2,t 1 ,..., FX k ,t 1 ) , equation (11) is defined as a Markov multi-variant fuzzy
process.
The model algorithm for the Markov fuzzy time series model is presented as
follows:
Step1. Define the universe of discourse.
Step2. Maximize fuzzy membership function.
Step3. Decide the order of the fuzzy auto-regressive model.
Step4. Compute Markov fuzzy relation matrix.
Step5. Forecast.
III. Empirical results
All the research data were gathered from the AREMOS economic statistics
database. The selected data were the amount of Taiwan export during the time period
between January 1990 and 30 March 2002. A total of 147 data points were collected
and shown in the Fig 1. The fuzzy time series model discussed includes 3 types, the
Factors model, the Heuristic model and the Markov model. For comparison purpose,
the ARIMA model was used as a reference point to contrast the forecast predictions,
and the mean square error (MSE) was used to assess the performance ability of the
models.

3.1 The ARIMA model with the whole data

The result of ARIMA model is obtained as the following:
X 11t = t 0.79 t 1 + 0.308 t 12 0.39 t 13 X 11 t = log X 1 t
Based on the MSE of ARIMA (Table 2), it could be interpreted that the model is a
seasonal moving average model, meaning that the amount of export is affected by the
residual error of lag 1, lag 12 and lag 13. The amount of Taiwan export could be
mainly attributable to the external Factors, especially when the coefficient of the error
term at the previous period increases to 79%, reflecting the influences of the external
Factors. This impact reduces with time and the export amounts after the lag 12 or 13
period do not affect current export amount anymore. Noticeably, the export amount
8

reaches its peak at the end of year. In general, the export amounts grow gradually annually,
except for the period from September 2000 to May 2001.

3.2 The fuzzy time series model with whole data

After the Factor model algorithm computation, the MSE of Factors model shown in
table 1.

This study compares the MSE value of two-month window basis w = 2 ,

three-month window basis w = 3 , four-month window basis w = 4 to examine

the forecasting ability of the models. The results indicate that the MSE forecasting
value of two-month window basis is the smallest, followed by the three-month
window basis. It also shows a trend that the more lag periods back, the larger the
MSEs.
Finally, according to the Markov model algorithm, this study also compares the
forecasting values of one-period ahead ( t = 1 ), two-period ahead ( t = 2 ), and
three-period ahead ( t = 3 ). The results show that the MSE value of two-period
ahead is the smallest. On the contrary, the Markov model shows that the more lag
periods back, the smaller the MSEs.
In summary, the results suggest that the ARIMA model has better forecasting power
with the smallest MSE among the four methods. However, the results are calculated
using a long-term period, causing the ARIMA to capture easily the trend for the
amount of export. In the fuzzy time series, the Heuristic model has a smaller
forecasting error which may distinguish the superiority of Heuristic model from the
others when using historical data to forecast the future trend.

3.3 The ARIMA models and Fuzzy Time Series models with partition data

To further investigate whether the differences in the prediction of models were

related to the length of time span, this study has formed three sets of data separated by
different time periods. First, from the test period between January 1995 and March
2002, the data on export values consist of a total of 87 files shown as Fig. 2:
Second, from the test period between January 1998 and March 2002, a total of 51 files
are collected as in Fig. 3:
Third, from the test period between January 2000 and March 2002, a total of 27 files
are found and shown in Fig. 4

3.4 Comparing the results of the four models

Table 2 depicts the prediction results for the three sample test periods estimated
from the four different models. According to the table, the ARIMA model shows the
forecast advantage, with little prediction errors when the test period is lengthy. In
contract, when the test period is relatively shorter, the fuzzy time series model has
been proved to be more effective than the ARIMA model. Overall, the results show
the Heuristic model has the lowest prediction error, followed by the Markov model.

IV. Conclusion
Previous research reveals that the export values of Taiwan is predominately
influenced by the external Factors but not past experiences. However, the export
values show an increasing trend generally. This study recommends that the results
for the testing periods are influenced by the first, twelfth and thirteenth period prior to
the test period, with the first period having the greatest power. The results show that
as the testing period extends, the lower of the MSE is found especially in the time
series ARIMA model. It suggests that the ARIMA model can compensate and balance
the fluctuating export values over time. In contrast, the shorter the testing period, the
more effectiveness is the fuzzy time series model over the ARIMA model.
In terms of research method, as the research period becomes shorter, the fuzzy time
series model depicts a declining trend of the MSE. The prediction results prove that
the Heuristic model is particularly effective and efficient. Its calculation method is
simple and easy to estimate. Therefore, when confronting with the lack of data
observations or when requiring more timely forecasts, the fuzzy time series model
could be a very useful and practical prediction tool to apply.

Reference
1.Chen, S. ,Forecasting enrollments based on fuzzy time series,Fuzzy Sets and Systems
,18,311-3191996.
2.Chen, S.and J.Hwang., Temperature prediction using fuzzy time series,IEEE Transactions
on System ,Man and Cybernetic ,-Part B: Cybernetics ,VoL.30 ,No.2 , 263-375
2000.
3.Own C.M.and P.T. Tu. , Forecasting fuzzy times on a heuristic high-order model,
1

Cybernetics and Systems:An International Journal ,36 ,705-717(2005).

4.Hwang, J., Chen, S., Lee, C. , Handling forecasting problems using fuzzy time series,
Fuzzy Sets and Systems,100,217-2281998.
5.Huarng, K. , Heuristic models of fuzzy time series for forecasting, Fuzzy Sets and
Systems,123,369-3862001.
6..Lee,L.,Wang,L.,Chen,S., Handling forecasting problems based on two-factors high-order
fuzzy time series, IEEE Transactions on fuzzy System,14(3), 468-477(2006).
7.Lee, L., Wang, L., Chen, S., Temperature prediction and TAIFEX forecasting based on
high-order fuzzy logical relationships and genetic simulated annealing techniques,
Expert System with Applications, 34,328-336(2008).
8.Sullivan. J.and W.Woodall., A comparision of fuzzy forecasting and Markov modeling,
Fuzzy Sets and Systems,64,279-293(1994.
9.Song, Q. and B. S. Chissom., Forecasting enrollments with fuzzy tmie series-Part I,
Fuzzy Sets and Systems,54,1-91993a.
10.Song, Q. and B. S. Chissom., Fuzzy time series and its models, Fuzzy Sets and
Systems,54,269-2771993b.
11.Song, Q. and B. S. Chisso., Forecasting enrollments with fuzzy time series Part II,
Fuzzy Sets and Systems,62,1-81994.
12.Song,Q. and R.P. Leland ., Adaptive learning defuzzification techniques and applications,
Fuzzy Sets and Systems ,81(3) ,321-329(1996).

200000

100000

M
e
a
nE
X
P
O
R
T

500000

400000

300000

0
.0
01
02 0
20 5.0
0
01 0
20 9.0
0
00 0
20 1.0
0
00 0
20 5.0
0
99 0
19 9.0
0
98 0
19 1.0
0
98 0
19 5.0
0
97 0
19 .0
09
96 0
19 1.0
0
96 0
19 .0
05
95 0
19 9.0
0
94 0
19 1.0
0
94 0
19 5.0
0
93 0
19 9.0
0
92 0
19 1.0
0
92 0
19 5.0
0
91 0
19 9.0
0
90 0
19 1.0
0
90
19

DATE

Figure 1 :Taiwan export amount from Jan. 1990 to March 2002

200000

100000

M
e
a
n
E
X
P
O
R
T

500000

400000

300000

0
.0
02
02
20 00
.
09
01
20 .00
04
01
20 00
.
11
00
20 .00
06
00
20 00
.
01
00
20 .00
08
99
19 00
.
03
99
19 .00
10
98
19 00
.
05
98
19 .00
12
97
19 00
.
07
97
19 .00
02
97
19 00
.
09
96
19 .00
04
96
19 00
.
11
95
19 .00
06
95
19 00
.
01
95
19

DATE

Figure 2: Taiwan export amount from May 1995 to March 2002

300000

200000

M
e
a
n
E
X
P
O
R
T

500000

400000

0
.0
01
02
20
0
.0
10
01
0
20
.0
07
01
20
0
.0
04
01
0
20
.0
01
01
20
0
.0
10
00
0
20
.0
07
00
20
0
.0
04
00
0
20
.0
01
00
20
0
.0
10
99
19
0
.0
07
99
0
19
.0
04
99
19
0
.0
01
99
0
19
.0
10
98
19
0
.0
07
98
0
19
.0
04
98
19
0
.0
01
98
19

DATE

Figure 3: Taiwan export amount from Jan. 1998 to March 2002

300000

200000

M
e
a
nE
X
P
O
R
T

500000

400000

0
.0
03
02
20
0
.0
01
02
20
0
.0
11
01
20
0
.0
09
01
20
0
.0
07
01
20
0
.0
05
01
20
0
.0
03
01
20
0
.0
01
01
20
0
.0
11
00
20
0
.0
09
00
20
0
.0
07
00
20
0
.0
05
00
20
0
.0
03
00
20
0
.0
01
00
20

DATE

Figure 4 :Taiwan export amount from Jan. 2000 to March 2002

Table 1: 4 Models MSE Result Comparison

ARIMA

Factor model

Heuristic model

w=2
6.14108***
Note:

1.58109*

Markov model

t=2
1.01109**

***smallest MSE value; **2nd smallest MSE; *3rd smallest MSE

2.56109

Table 2. The MSE results of the three sample test periods measured from four different models
Period

ARIMA

199501-200203

Factor Model

9.28108***

w=2

Heuristic Model

2.25109*

1.16109**

Markov model

t=2

2.40109

199801-200203

1.34109*

2.56109

5.86108***

1.14109**

200001-200203

1.66109*

2.36109

5.47108***

1.37109**

Note:

***smallest MSE value; **2nd smallest MSE; *3rd smallest MSE

Appendix
Appendix 1 The whole data 199001-200203
Date Export ARIMA

Factor(w=2)

Factor(w=3) Factor(w=4) Factor(w=5) Factor(w=6)

Heuristic Markov

199001

128808

199002

131563

129728.6188

171340

129108

199003

137805

131045.517

171340

101863

199004

150366

133345.6622

153105

171340

138105

199005

148390

137654.1969

165666

165666

150070

120666

199006

144279

140786.5229

148690

148690

148690

150070

148690

199007

169539

142514.6385

129579

129579

129579

129579

171340

144579

199008

157044

148731.8055

169839

169839

184839

184839

184839

150070

169839

199009

151931

151478.6546

157344

142344

142344

142344

142344

150070

157344

199010

168756

152664.808

137231

137231

137231

137231

137231

171340

152231

199011

165632

156952.7369

169056

169056

184056

184056

184056

150070

169056

199012

148670

159835.4296

165932

150932

150932

150932

150932

150070

180932

199101

168382

158609.423

133970

133970

133970

133970

133970

171340

148970

199102

132881

162413.365

168682

168682

183682

183682

183682

150070

183682

199103

155051

158577.7043

118181

103181

103181

103181

103181

171340

103181

199104

153698

161748.8334

140351

170351

170351

170351

170351

150070

155351

199105

185328

157440.0453

153998

138998

138998

138998

138998

171340

153998

199106

187228

160427.9161

200628

215628

215628

215628

215628

235150

185628

199107

191500

174241.9274

217528

202528

202528

202528

202528

235150

187528

199108

164614

170024.8701

206800

206800

206800

206800

206800

171340

206800

199109

184603

166691.4871

164914

164914

164914

149914

149914

171340

164914

199110

175562

176592.4828

184903

199903

199903

199903

199903

171340

184903

199111

182891

173685.2444

175862

160862

160862

160862

160862

235150

130862

199112

159047

169312.2632

183191

198191

198191

198191

198191

171340

198191

199201

176723

176398.3379

159347

144347

144347

144347

144347

171340

114347

199202

130299

163243.7889

177023

192023

192023

192023

192023

171340

207023

199203

174902

168619.504

115599

100599

100599

100599

100599

171340

100599

199204

177325

169922.8858

190202

205202

205202

205202

205202

235150

175202

199205

174956

185183.0362

207625

207625

192625

192625

192625

171340

177625

199206

170289

181407.2894

175256

175256

160256

160256

160256

171340

175256

199207

169453

174817.8756

155589

155589

155589

155589

155589

150070

140589

199208

182303

166958.4069

154753

154753

154753

154753

154753

171340

169753

199209

163075

178903.172

182603

182603

182603

197603

197603

171340

197603

199210

177243

169600.3388

163375

148375

148375

148375

148375

171340

163375

199211

189475

175518.5475

177543

192543

192543

192543

192543

235150

177543

199212

161920

172646.253

204775

204775

204775

204775

204775

171340

204775

199301

161192

175957.8928

162220

162220

147220

147220

147220

150070

162220

199302

161821

162299.7154

146492

146492

146492

146492

146492

171340

131492

199303

191207

180412.2881

162121

162121

177121

177121

177121

213880

132121

199304

180204

183670.7442

206507

206507

206507

206507

206507

171340

191507

199305

206636

178082.1073

180504

180504

165504

165504

165504

235150

180504

199306

183095

185423.9307

206936

221936

221936

221936

221936

171340

206936

199307

190929

189113.0002

183395

168395

168395

168395

168395

235150

153395

199308

204520

198471.3428

191229

206229

206229

206229

206229

235150

191229

199309

172943

188886.2007

219820

219820

219820

219820

219820

171340

219820

199310

200060

196359.0205

158243

158243

143243

143243

143243

235150

173243

199311

201467

199785.5859

185360

215360

215360

215360

215360

235150

200360

199312

184958

191817.5617

216767

216767

216767

216767

216767

171340

216767

199401

195583

191396.027

185258

185258

170258

170258

170258

235150

185258

199402

151291

200185.5569

195883

210883

210883

210883

210883

171340

210883

199403

188766

194055.601

136591

121591

121591

121591

106591

171340

151591

199404

196141

188867.313

204066

219066

219066

219066

219066

213880

189066

199405

223982

202177.2365

226441

226441

211441

211441

211441

235150

166441

199406

194726

195457.8433

239282

239282

239282

239282

239282

171340

239282

199407

209555

198272.9707

195026

195026

195026

180026

180026

171340

195026

199408

211785

203089.0842

209855

224855

224855

224855

224855

235150

209855

199409

198917

198451.4967

227085

227085

227085

227085

227085

171340

212085

199410

229080

208167.8491

199217

199217

184217

184217

184217

213880

199217

199411

246252

212798.2899

229380

244380

244380

244380

244380

291870

229380

199412

209933

217662.2873

261552

261552

261552

261552

261552

171340

261552

199501

193994

222214.8532

195233

195233

180233

180233

180233

171340

225233

199502

223822

198035.1858

164294

179294

179294

179294

179294

235150

194294

199503

237050

226205.6907

224122

224122

239122

239122

239122

291870

224122

199504

246134

235225.5218

252350

252350

252350

252350

252350

291870

192350

199505

255020

243166.0932

261434

261434

261434

261434

261434

291870

246434

199506

221121

237352.1067

270320

270320

270320

270320

270320

171340

255320

199507

263210

240025.3928

206421

206421

206421

206421

191421

291870

191421

199508

265054

244213.9587

278510

293510

293510

293510

293510

341500

233510

199509

239023

246249.3698

295354

295354

280354

280354

280354

235150

280354

199510

283433

253647.3004

239323

239323

224323

224323

224323

291870

239323

199511

269456

263313.3377

298733

313733

328733

328733

328733

235150

253733

199512

252261

248923.2761

284756

254756

254756

254756

254756

235150

284756

199601

291615

244556.0977

237561

237561

237561

237561

237561

291870

252561

199602

206925

279567.767

306915

306915

336915

336915

336915

235150

321915

199603

247100

256068.4319

222225

162225

162225

162225

162225

235150

207225

199604

273929

254918.7128

262400

292400

292400

292400

292400

291870

247400

199605

259226

259756.5175

304229

304229

304229

289229

289229

341500

244229

199606

267381

251173.6973

259526

259526

259526

259526

244526

341500

259526

199607

261020

271007.019

267681

282681

282681

282681

282681

235150

267681

199608

266508

267961.8772

261320

246320

246320

246320

246320

341500

261320

199609

269353

258719.5667

266808

281808

281808

281808

281808

341500

266808

199610

269175

275047.7506

284653

284653

284653

284653

284653

235150

269653

199611

282082

265986.278

269475

269475

254475

254475

254475

341500

269475

199612

282311

269817.2076

282382

297382

297382

297382

297382

341500

282382

199701

283520

288024.2624

282611

267611

267611

267611

267611

341500

282611

199702

212689

246126.2759

283820

298820

298820

298820

298820

235150

313820

199703

292756

267613.7531

227989

167989

167989

167989

167989

235150

167989

199704

270694

284617.0866

308056

338056

338056

338056

338056

235150

308056

199705

280615

275708.6579

285994

225994

270994

255994

255994

341500

270994

199706

276885

284332.6118

280915

280915

280915

295915

295915

235150

280915

199707

297719

274772.8403

277185

262185

262185

262185

262185

341500

247185

199708

270702

285015.8296

298019

313019

313019

313019

313019

235150

313019

199709

310363

288197.1288

271002

256002

256002

256002

256002

341500

271002

199710

296202

288197.1288

325663

340663

340663

340663

340663

277690

355663

199711

350163

299718.8058

311502

281502

296502

296502

296502

341500

281502

199712

339377

308846.6023

365463

395463

395463

395463

395463

320230

320463

199801

252411

310084.4628

354677

324677

324677

324677

324677

277690

414677

199802

288227

288053.0663

207711

207711

207711

207711

207711

291870

297711

199803

341595

315779.6692

303527

303527

333527

333527

333527

341500

273527

199804

299105

307306.2235

386895

386895

386895

386895

386895

277690

356895

199805

309605

315590.2582

314405

314405

239405

254405

254405

341500

269405

199806

311645

311888.1784

294905

339905

339905

339905

324905

362770

279905

199807

307267

325266.4592

326945

326945

326945

326945

326945

277690

311945

199808

331932

308908.3778

307567

307567

292567

292567

292567

362770

277567

199809

327980

329588.4388

332232

347232

347232

347232

347232

277690

287232

199810

291567

324876.3736

328280

313280

313280

313280

313280

277690

373280

199811

339896

332169.2808

306867

306867

306867

306867

306867

341500

276867

199812

292039

325624.4492

355196

355196

370196

370196

370196

277690

370196

199901

315110

290221.586

307339

247339

247339

247339

247339

341500

292339

199902

244105

322191.0592

300410

345410

345410

330410

330410

277690

330410

199903

343620

313984.8452

259405

199405

199405

199405

199405

291870

199405

199904

301524

309836.4944

358920

358920

358920

418920

418920

320230

358920

199905

347390

311763.4481

316824

241824

241824

256824

256824

362770

286824

199906

320562

323385.3742

362690

422690

422690

377690

377690

320230

362690

199907

321293

319559.895

335862

290862

290862

290862

305862

362770

320862

199908

343608

336818.6723

321593

336593

336593

336593

336593

362770

321593

199909

303197

330710.9467

358908

358908

358908

358908

358908

320230

358908

199910

362954

317108.7329

288497

288497

273497

273497

258497

362770

288497

199911

365463

344931.1118

378254

393254

393254

393254

408254

384040

363254

199912

348619

336953.4267

395763

395763

380763

380763

380763

320230

380763

200001

367435

365711.4816

348919

348919

333919

333919

333919

384040

303919

200002

277702

327846.241

367735

382735

382735

382735

382735

320230

397735

200003

366727

365674.9123

293002

293002

293002

293002

293002

341500

233002

200004

373491

352815.3108

382027

412027

412027

412027

427027

384040

367027

200005

402271

375645.0648

418791

418791

388791

388791

388791

384040

373791

200006

381285

366846.9463

417571

417571

417571

417571

417571

362770

417571

200007

417615

374295.1738

381585

381585

381585

381585

381585

384040

351585

200008

395466

387085.1602

432915

447915

447915

462915

462915

362770

432915

200009

401747

378321.6354

410766

380766

380766

380766

380766

405310

365766

200010

426525

415983.136

402047

417047

417047

417047

417047

405310

372047

200011

407275

406687.3551

441825

441825

441825

441825

441825

362770

426825

200012

398762

404780.4094

407575

407575

392575

392575

392575

362770

407575

200101

323287

401675.5692

384062

384062

384062

384062

384062

362770

429062

200102

327601

367030.4156

278587

278587

278587

278587

278587

362770

323587

200103

379680

385694.1589

342901

327901

327901

342901

342901

362770

312901

200104

356122

393603.7612

409980

409980

409980

409980

409980

320230

394980

200105

333408

387937.685

371422

356422

356422

341422

341422

320230

356422

200106

350486

373211.2902

318708

318708

318708

318708

318708

362770

303708

200107

335950

378056.9029

350786

350786

365786

365786

365786

320230

320786

200108

326822

358469.9091

336250

321250

321250

321250

321250

277690

336250

200109

305762

357896.8159

312122

312122

312122

312122

312122

277690

342122

200110

394626

343588.5002

291062

291062

291062

291062

291062

362770

261062

200111

350545

352709.482

409926

409926

409926

439926

439926

362770

409926

200112

353455

352956.4651

365845

320845

320845

320845

320845

384040

320845

200201

338326

334602.9888

338755

368755

353755

353755

353755

320230

353755

200202

281746

355009.5608

338626

323626

323626

323626

323626

277690

368626

200203

400749

354761.1411

252046

252046

237046

237046

237046

341500

237046

MSE

614182535.1 1579441508 2275912692 2451959259 2484624028 2552875331 1013539018 2557303778

Appendix 2:Partition data 199501-200203

DATE

Export

ARIMA

Factor(w=2) Factor(w=3) Factor(w=4) Factor(w=5) Factor(w=6) Factor(w=7)

Heuristic Markov

199501

193994

277153.3

199502

223822

195262.5

199503

237050

204270.4

199504

246134

214443.2

266597

199505

255020

224425.7

260771

260771

199506

221121

234192.8

269657

269657

269657

199507

263210

231932.1

205938

205938

205938

205938

199508

265054

241953.3

277847

292757

292757

292757

307667

199509

239023

249896

294601

294601

279691

279691

279691

279691

199510

283433

248376.2

238750

238750

223840

223840

223840

223840 288113.3

243820 277153.3
243820 277153.3
288113.3 277153.3
288113.3 277153.3
210600 277153.3
243820 277153.3
343480 277153.3
243820 277153.3
310430

199511

269456

259522.8

298070

312980

327890

327890

327890

327890

243820 277153.3

199512

252261

263998.8

284093

254273

254273

254273

254273

254273

243820 277153.3

199601

291615

262314.6

237078

237078

237078

237078

237078

237078 288113.3 310373.3

199602

206925

285500.8

306252

306252

336072

336072

336072

336072

243820 277153.3

199603

247100

261921.5

221562

161922

161922

161922

161922

161922

243820 277153.3

199604

273929

256606.7

261737

291557

291557

291557

291557

291557 288113.3 277153.3

199605

259226

260563

303476

303476

303476

288566

288566

288566

199606

267381

243896.7

258953

258953

258953

258953

244043

244043 288113.3 277153.3

199607

261020

269574.5

267108

282018

282018

282018

282018

282018

199608

266508

264025.2

260747

245837

245837

245837

245837

245837

343480 277153.3

199609

269353

252963.4

266235

281145

281145

281145

281145

281145

343480 277153.3

199610

269175

276398.8

283990

283990

283990

283990

283990

283990

243820

199611

282082

265507.9

283812

283812

283812

283812

283812

283812

343480 310373.3

199612

282311

263550.4

296719

296719

296719

296719

296719

296719

343840 310373.3

199701

283520

285900.7

296948

296948

296948

296948

296948

296948

343480

310430

199702

212689

245659.1

298157

298157

298157

298157

298157

298157

243820

377040

199703

292756

264342.3

227326

227326

227326

227326

227326

227326

243820

343650

199704

270694

286817.1

307393

337213

337213

337213

337213

337213

243820

377040

199705

280615

276233

285331

225691

270421

270421

270421

270421

343480

343650

199706

276885

289091.9

280342

280342

280342

295252

295252

295252

243820

343650

199707

297719

273977.2

276612

261702

261702

261702

261702

261702

343480

343650

199708

270702

286903.2

297446

312356

312356

312356

312356

312356

277040

377040

199709

310363

289381.2

270429

255519

255519

255519

255519

255519

343480

343650

199710

296202

287046.6

325000

339910

339910

339910

339910

354820

277040

376700

199711

350163

300980.3

310839

281019

295929

295929

295929

295929

360090

343480

199712

339377

316253.7

364800

394620

394620

394620

394620

394620

293650 299186.7

243820

243820

377040

377040

377040

199801

252411

315022.7

354014

324194

324194

324194

324194

324194

293650

360260

199802

288227

283991.6

207408

207408

207408

207408

207408

207408 288113.3

326870

199803

341595

315148.7

302864

302864

332684

332684

332684

332684

343480 299186.7

199804

299105

305284.7

386052

386052

386052

386052

386052

386052

293650

326870

199805

309605

314361.9

313742

313742

239192

254102

254102

254102

360090

360175

199806

311645

308352.8

294422

339152

339152

339152

324242

324242

360090 343593.3

199807

307267

326113.3

326282

326282

326282

326282

326282

326282

277040

199808

331932

305926.5

306994

306994

292084

292084

292084

292084

360090 343593.3

199809

327980

330578.7

331659

346569

346569

346569

346569

346569

293650 343593.3

199810

291567

326472.2

327707

312797

312797

312797

312797

312797

293650

360175

326870

199811

339896

331571.9

261474

261474

261474

261474

261474

246564

343480 299186.7

199812

292039

324421.9

354533

354533

369443

369443

369443

369443

293650

199901

315110

284987.3

306676

247036

247036

247036

247036

247036

343480 299186.7

199902

244105

324032.8

299927

344657

344657

329747

329747

329747

277040 321446.7

199903

343620

307890.7

258742

199102

199102

199102

199102

199102 288113.3 343593.3

199904

301524

307367.7

358257

358257

358257

417897

417897

417897

293650

326870

199905

347390

308537.9

316161

241611

241611

256521

256521

256521

360090

343480

199906

320562

324324.6

362027

421667

421667

376937

376937

391847

293650 299186.7

199907

321293

317235.6

335199

290469

290469

290469

305379

305379

360090 299186.7

199908

343608

337763.1

321020

335930

335930

335930

335930

335930

360090 299186.7

199909

303197

329786.3

358245

358245

358245

358245

358245

358245

293650

326870

199910

362954

312418.8

288014

288014

273104

273104

258194

258194

360090

343480

199911

365463

346209.7

377591

407411

407411

407411

422321

422321

393310

360090

199912

348619

337594.2

409920

380100

380100

380100

380100

380100

310260

360090

200001

367435

370459.7

348346

348346

333436

333436

333436

333436

360090

360090

200002

277702

323417.7

367162

382072

382072

382072

382072

382072

310260

376700

200003

366727

362797.5

292339

232699

232699

232699

217789

217789

343480

360090

326870

200004

373491

348258.4

381364

411184

411184

411184

441004

441004

393310

360090

200005

402271

375682.6

417948

417948

388128

388128

388128

388128

393310

376700

200006

381285

366223.8

416908

416908

416908

416908

416908

416908

360090

376700

200007

417615

376284.2

381012

381012

381012

381012

381012

366102

393310

376700

200008

395466

390467.5

432252

447162

447162

462072

462072

462072

360090

376700

200009

401747

380256

410103

380283

380283

380283

380283

380283

409920

376700

200010

426525

423835.2

401474

416384

416384

416384

416384

416384

409920 354553.3

200011

407275

410405.1

441162

441162

441162

441162

441162

441162

360090

376700

200012

398762

407990.8

407002

407002

392092

392092

392092

392092

360090

376700

200101

323287

401675.6

383579

383579

383579

383579

383579

383579

360090 299186.7

200102

327601

361457.6

278284

278284

278284

278284

278284

278284

360090 299186.7

200103

379680

377414.8

342238

327328

327328

342238

342238

342238

360090

376700

200104

356122

388364.7

409227

409227

409227

409227

409227

409227

310260

343480

200105

333408

380027.9

370759

355849

355849

340939

340939

340939

293650 343593.3

200106

350486

363851.1

318225

318225

318225

318225

318225

318225

360090

360260

200107

335950

370015.4

350213

350213

365123

365123

365123

365123

293650 343593.3

200108

326822

349234.9

335677

320767

320767

320767

320767

320767

293650 343593.3

200109

305762

349969

311639

311639

311639

311639

311639

311639

277040

360175

200110

394626

332468.4

290579

290579

290579

290579

290579

290579

360090

376700

200111

350545

349409.5

409263

409263

409263

439083

439083

439083

360090

360260

200112

353455

350249.1

365182

320452

320452

320452

320452

320452

409920

360260

200201

338326

330347.4

338272

368092

353182

353182

353182

368092

293650 343593.3

200202

281746

352497.9

338053

323143

323143

323143

323143

323143

293650

360175

200203

400749

347041.6

251653

251653

236743

236743

236743

236743

343480

376700

MSE

9.28E+08 2.25E+09 3.25E+09 3.51E+09 3.52E+09 3.67E+09 3.79E+09 1.16E+09 2.4E+09

Appendix 3: Partition data 199801-200203

Date

Export

ARIMA

Factor(w=2) Factor(w=3) Factor(w=4) Factor(w=5) Factor(w=6) Factor(w=7)

Heuristic Markov

199801

252411

309255

199802

288227

253393.8

257135

322285

199803

341595

268632.6

340527

348345

199804

299105

298582

386052

309255

309255

199805

309605

299958.7

313742

313742

348345

309255

199806

311645

305193.1

294422

339152

339152

348345

335315

199807

307267

309093.8

326282

326282

326282

326282

283195

335315

199808

331932

309526.8

306994

306994

292084

292084

292084

348345

348345

199809

327980

320071.6

331659

346569

346569

346569

346569

346569

309255

335315

199810

291567

324649

327707

312797

312797

312797

312797

312797

309255

322285

199811

339896

311389.6

261474

261474

261474

261474

261474

246564

340527

335315

199812

292039

324421.9

354533

354533

369443

369443

369443

369443

309255

322285

199901

315110

311483

306676

247036

247036

247036

247036

247036

340527 344001.7

199902

244105

320744.5

299927

344657

344657

329747

329747

329747

283195

335315

199903

343620

297955.7

258742

258742

258742

258742

258742

258742 291881.7

309255

199904

301524

317838.9

358257

358257

358257

417897

417897

417897

309255

322285

199905

347390

313576.9

316161

241611

241611

256521

256521

256521

348345

335315

199906

320562

329852.2

362027

421667

421667

376937

376937

391847

309255 344001.7

199907

321293

326831.5

335199

290469

290469

290469

305379

305379

348345 344001.7

199908

343608

329390.7

321020

335930

335930

335930

335930

335930

348345

309255

199909

303197

337932

358245

358245

358245

358245

358245

358245

309255

361375

199910

362954

318507.1

288014

288014

273104

273104

258194

258194

348345

413495

199911

365463

342696.3

377591

407411

407411

407411

422321

422321

387435

413495

199912

348619

347666.9

409920

380100

380100

380100

380100

380100 326628.3

361375

200001

367435

350039.1

348346

348346

333436

333436

333436

333436

387435

387435

200002

277702

343451.1

367162

382072

382072

382072

382072

382072 326628.3

387435

200003

366727

322320

292339

232699

232699

232699

232699

232699

340527

387435

200004

373491

338913.4

381364

411184

411184

411184

441004

441004

387435

387435

200005

402271

360338.8

417948

417948

388128

388128

388128

388128

387435

413495

200006

381285

377301.5

416908

416908

416908

416908

416908

416908

400465

387435

200007

417615

379420.4

381012

381012

381012

381012

381012

366102

400465

413495

200008

395466

399352.6

432252

447162

447162

462072

462072

462072

400465

413495

200009

401747

392424.7

410103

380283

380283

380283

380283

380283

400465

413495

200010

426525

406240.2

401474

416384

416384

416384

416384

416384

413495

387435

200011

407275

420584.2

441162

441162

441162

441162

441162

441162

400465

387435

200012

398762

416690.9

421912

421912

407002

407002

407002

407002

400465

348345

200101

323287

413742.9

383579

383579

383579

383579

383579

383579

348345 344001.7

200102

327601

361674.5

278284

278284

278284

278284

278284

278284

374405 344001.7

200103

379680

355364.7

342238

327328

327328

342238

342238

342238

374405

348345

200104

356122

372614.6

409227

409227

409227

409227

409227

409227

348345

348345

200105

333408

373435.3

370759

355849

355849

340939

340939

340939 326628.3

309255

200106

350486

357896.8

318225

318225

333135

333135

333135

333135

200107

335950

361566

350213

350213

365123

365123

365123

365123 326628.3 326628.3

200108

326822

351301.5

335677

320767

320767

320767

320767

320767

309255

335315

200109

305762

343313.7

311639

311639

311639

311639

311639

311639

309255

335315

200110

394626

330677.9

290579

290579

290579

290579

290579

290579

348345

348345

374405 326628.3

200111

350545

355969.4

409263

409263

409263

439083

439083

439083

348345 326628.3

200112

353455

352533.2

365182

320452

320452

320452

320452

320452

387435 326628.3

200201

338326

340646.3

338272

368092

353182

353182

353182

368092 326628.3 326628.3

200202

281746

335641.9

338053

323143

323143

323143

323143

323143

309255

335315

200203

400749

316158.8

251653

251653

236743

236743

236743

236743

340527

413495

MSE

1.34E+09 2.57E+09 3.48E+09 3.77E+09 3.88E+09 4.11E+09 4.32E+09 5.86E+08 1.14E+09

Appendix 4 :Partition data 20001-200203

Date

Export

ARIMA

Factor(w=2) Factor(w=3) Factor(w=4) Factor(w=5) Factor(w=6) Factor(w=7)

Heuristic Markov

200001

367435

200002

277702

366663.6

200003

366727

363523.8

200004

373491

362870

381364

200005

402271

362362.4

417948

417948

200006

381285

362507.3

416908

416908

416908

200007

417615

362217.4

381012

381012

381012

381012

200008

395466

362688.6

432252

447162

447162

462072

462072

200009

401747

362652.4

410103

380283

380283

380283

380283

380283

405266 415896.5

200010

426525

362797.5

401474

416384

416384

416384

416384

416384

405266 415896.5

200011

407275

363451.1

441162

441162

441162

441162

441162

441162

405266 394635.5

200012

398762

363669.2

407002

407002

392092

392092

392092

392092

405266 394635.5

200101

323287

363705.6

383579

383579

383579

383579

383579

383579 352113.5 330852.5

200102

327601

364324.4

337924

337924

337924

337924

337924

337924 352113.5 330852.5

200103

379680

362616.1

342238

327328

327328

342238

342238

342238 352113.5 394635.5

200104

356122

362000.2

409227

409227

409227

409227

409227

409227 337939.5

341483

200105

333408

360230.7

370759

355849

355849

340939

340939

340939 330852.5

341483

200106

350486

358398.2

318225

318225

333135

333135

333135

333135 352113.5

341483

200107

335950

356289.9

350213

350213

365123

365123

365123

365123 330852.5

341483

200108

326822

354335.7

335677

320767

320767

320767

320767

320767 309591.5

341483

200109

305762

352075.2

311639

311639

311639

311639

311639

311639 309591.5 394635.5

200110

394626

348850.9

290579

290579

290579

290579

290579

290579 352113.5

341483

200111

350545

348223.6

409263

409263

409263

439083

439083

439083 352113.5

341483

200112

353455

346798.8

365182

320452

320452

320452

320452

320452 352113.5

341483

200201

338326

347215.2

338272

368092

353182

353182

353182

368092 330852.5

341483

200202

281746

347145.8

338053

323143

323143

323143

323143

323143 309591.5 394635.5

200203

400749

344379.7

251653

251653

236743

236743

236743

236743 352113.5 394635.5

MSE

368059.3
337939.5

384005

352113.5 368059.3
394635.5 368059.3
394635.5

405266

352113.5 394635.5
394635.5 415896.5
405266 415896.5

1.66E+09 2.36E+09 2.55E+09 2.83E+09 3.11E+09 5.34E+08 3.17E+09 5.47E+08 1.37E+09