Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів
The paper considers the problem of short- and middle-term forecasting in the financial sphere. To solve this problem, a hybrid system of computational intelligence based on the group method of data handling (GMDH) and bagging, as well as an algorithm for its training, is proposed. The odd stacks of...
Gespeichert in:
| Datum: | 2024 |
|---|---|
| Hauptverfasser: | , , , |
| Format: | Artikel |
| Sprache: | Englisch |
| Veröffentlicht: |
The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute"
2024
|
| Schlagworte: | |
| Online Zugang: | https://journal.iasa.kpi.ua/article/view/304428 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | System research and information technologies |
| Завантажити файл: |  |
Institution
System research and information technologies| _version_ | 1867334443907153920 |
|---|---|
| author | Bodyanskiy, Yevgeniy Kuzmenko, Oleksii Zaichenko, Helen Zaychenko, Yuriy |
| author_facet | Bodyanskiy, Yevgeniy Kuzmenko, Oleksii Zaichenko, Helen Zaychenko, Yuriy |
| author_institution_txt_mv | [
{
"author": "Yevgeniy Bodyanskiy",
"institution": "Kharkiv National University of Radio Electronics, Kharkiv"
},
{
"author": "Oleksii Kuzmenko",
"institution": "Educational and Research Institute for Applied System Analysis of the National Technical University of Ukraine \"Igor Sikorsky Kyiv Polytechnic Institute\", Kyiv"
},
{
"author": "Helen Zaichenko",
"institution": "Educational and Research Institute for Applied System Analysis of the National Technical University of Ukraine \"Igor Sikorsky Kyiv Polytechnic Institute\", Kyiv"
},
{
"author": "Yuriy Zaychenko",
"institution": "Educational and Research Institute for Applied System Analysis of the National Technical University of Ukraine \"Igor Sikorsky Kyiv Polytechnic Institute\", Kyiv"
}
] |
| author_sort | Bodyanskiy, Yevgeniy |
| baseUrl_str | http://journal.iasa.kpi.ua/oai |
| collection | OJS |
| datestamp_date | 2024-05-23T07:09:36Z |
| description | The paper considers the problem of short- and middle-term forecasting in the financial sphere. To solve this problem, a hybrid system of computational intelligence based on the group method of data handling (GMDH) and bagging, as well as an algorithm for its training, is proposed. The odd stacks of the hybrid system are formed by ensembles of parallel connected subsystems. ARIMA and the GMDH-neo-fuzzy hybrid network were chosen as such subsystems. The proposed system does not require a large training data set, automatically determines the number of stacks during training, and provides online operation. The experimental investigations were conducted using the proposed hybrid system, as well as separately using ARIMA and GMDH-neo-fuzzy. The accuracy of the predictions obtained is compared, based on which the feasibility of using the proposed hybrid system is substantiated. |
| doi_str_mv | 10.20535/SRIT.2308-8893.2024.1.06 |
| first_indexed | 2025-07-17T10:28:30Z |
| format | Article |
| fulltext |
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko, 2024
Системні дослідження та інформаційні технології, 2024, № 1 75
TIДC
ТЕОРЕТИЧНІ ТА ПРИКЛАДНІ ПРОБЛЕМИ
ІНТЕЛЕКТУАЛЬНИХ СИСТЕМ ПІДТРИМАННЯ
ПРИЙНЯТТЯ РІШЕНЬ
UDC 519.925.51
DOI: 10.20535/SRIT.2308-8893.2024.1.06
HYBRID SYSTEM OF COMPUTATIONAL INTELLIGENCE
BASED ON BAGGING AND GROUP METHOD
OF DATA HANDLING
Ye. BODYANSKIY, O. KUZMENKO, He. ZAICHENKO, Yu. ZAYCHENKO
Abstract. The paper considers the problem of short- and middle-term forecasting in
the financial sphere. To solve this problem, a hybrid system of computational intel-
ligence based on the group method of data handling (GMDH) and bagging, as well
as an algorithm for its training, is proposed. The odd stacks of the hybrid system are
formed by ensembles of parallel connected subsystems. ARIMA and the GMDH-
neo-fuzzy hybrid network were chosen as such subsystems. The proposed system
does not require a large training data set, automatically determines the number of
stacks during training, and provides online operation. The experimental investiga-
tions were conducted using the proposed hybrid system, as well as separately using
ARIMA and GMDH-neo-fuzzy. The accuracy of the predictions obtained is compared,
based on which the feasibility of using the proposed hybrid system is substantiated.
Keywords: hybrid system, bagging, hybrid GMDH-neo-fuzzy network, ARIMA,
short- and middle-term forecasting.
INTRODUCTION
Today deep neural networks (DNN) are widely used for solving a large class of
Data mining problems and, above all, classification (pattern recognition, image
processing of various nature) and extrapolation (time series forecasting, natural
language text processing, fault detection), due to their universal approximating
properties, based mainly on the G. Cybenko [2] theorem. At the same time, DNNs
are rather cumbersome constructs that contain too many configurable synaptic
weights, in turn, requiring too much training data, which is not always available
for solving specific real-world problems. In addition, the learning process of DNN
requires quite a lot of time so that when solving Data Stream mining problems,
especially when processing nonstationary data streams, the use of these systems
encounters some difficulties.
It is possible to overcome these difficulties using ensemble methods [2–7],
while the ensemble can contain a variety of computational intelligence systems
the simplest type of elementary perceptron of F. Rosenblatt and N – Adaline [8]
to the most complex DNNs.
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko
ISSN 1681–6048 System Research & Information Technologies, 2024, № 1 76
The main problem when using the ensemble approach is to combine the re-
sults of individual members of this ensemble in order to obtain the optimal initial
result in the sensing of the accepted learning criterion. For this purpose, Bagging
procedures [9] can be used, minimizing the RMS error on the training sample,
modified to work online.
The resulting ensemble approach and bagging system may be too complex
from a computational point of view. To simplify its implementation, it should be
decomposed into a number of simpler subsystems, while such decomposition can
be implemented quite simply using the Group Method of Data Handling (GMDH)
[10; 11].
It is interesting to note that J. Schmidhuber [12] believes that just based on
GMDH the first deep learning systems were built. Subsequently, based on
GMDH, neural networks and neuro-fuzzy systems were proposed, which demon-
strated their accuracy and speed in solving a number of problems [13–17].
In our opinion, it is appropriate to introduce into consideration a hybrid sys-
tem of computational intelligence (HSCI), built based on an ensemble approach
and bagging, which would increase its architecture in the learning process based
on GMDH ideas, would not require significant amounts of training selections and
would be quite simple from a computational point of view.
ARCHITECTURE OF HSCI ON THE BASE OF BAGGING AND GMDH
In the Fig. 1 the architecture of the proposed system is presented.
The architecture of the system contains 2S sequentially-connected stacks,
while odd stacks are formed by ensembles of parallel-connected subsystems that
solve the same problem (recognition, prediction, etc.) and even ones are essen-
tially learning metamodels that generalize the output signals of ensembles and
form optimal results in the sense of the accepted criterion. The output signal of
the first metamodel is the generalized optimal signal )(1* ky and )1( n output
signals )(ˆ ]1[ kyi , )(1,,2,1 kni “best members of the ensemble”. At their core,
metamodels function as selection units in traditional GMDH systems, but not only
select the best results from the previous stack, but also form the optimal solution
based on these results.
Further, the output signals of the first metamodel are fed to the inputs of the
second ensemble, which is completely similar to the first. The outputs of the sec-
ond ensemble )(ˆ),..., (ˆ),(ˆ ]2[]2[
2
]2[
1 kykyky q come to the second metamodel, which
calculates the optimal signal )(]2*[ ky and )(ˆ)1( ]2[
kyn i “closest” to it. The last
S-th ensemble is similar to the first two, and the output of the last S-th metamodel
)(ˆ ]1[ kyq
)(ˆ ]2[ kyq
1
Fig. 1. Hybrid system of computational intelligence based on bagging and GMDH
x1(k)
x2(k)
xn(k) m
et
a
m
od
el
[
S
E
n
se
m
b
le
S
m
et
a
m
od
el
2
E
n
se
m
b
le
2
m
et
a
m
od
el
1
E
n
se
m
b
le
1
…
…
…
)(ˆ ]1[
1 ky
)(ˆ ]1[
1 ky )(ˆ ]1[
2 ky
)(ˆ ]1[
,1 kyk
..
)(]1[* ky )(ˆ ]2[
1 ky
)(ˆ ]2[
2 ky
.. ..
)(ˆ ]2[
1 ky
)(ˆ ]2[
,1 kyk
)(]2[* ky )(ˆ ][
1 ky s
)(ˆ ][
2 ky s
)(ˆ ][ ky s
q
.. .. ..
y+[s](k)
Hybrid system of computational intelligence based on bagging and group method …
Системні дослідження та інформаційні технології, 2024, № 1 77
is )(]*[ ky s , which exactly corresponds to a priori established requirements for the
quality of solving the problem under consideration.
Each of the ensembles contains q different computational intelligence sys-
tems that solve the same problem. There may still be simple neural networks such
as a single-layer perceptron, radial-basis neural network (RBFN), counterpropa-
gating neural network, etc., which do not use error backpropagation procedure for
training, neuro-fuzzy systems such as ANFIS, Wang-Mendel or Ta-
kagi-Sugeno-Kang type, wavelet-neuro systems, neo-fuzzy neurons and others,
the output signal of which linearly depends on the adapted parameters, which al-
lows to use optimal speed learning algorithms.
LEARNING HSCI BASED ON BAGGING AND GMDH
The input information, on the basis of which the system is configured, is a train-
ing selection of input signals );(),..., (),..., 2(), 1( Nxkxxx ),(,), (()( 1 kxkxkx i
nT
n Rkx ))(, and its corresponding scalar reference signals ),...,1(y
)( ),...,( Nyky . On the basis of these observations, the elements of the first en-
semble are tuned independently of each other, at the outputs of which q scalar
signals ,)(]2[ kyp qp ,..., 2,1 , are formed, which are conveniently represented in
the form of a vector T
qp kykykyky ))(ˆ,),(ˆ,),(ˆ()(ˆ ]1[]1[]1[
1
]1[ . These signals are
sent to the inputs of the first metamodel, at the outputs of which n sequences
)(ˆ,),(ˆ,),(ˆ),(ˆ ]1[
,1
]1[]1[
1
]1*[ kykykyky ni the main of which is )(ˆ ]1*[ ky while others
are auxiliary. The main signal of the metamodel )(]1*[ ky is the union of the out-
puts of all members of the ensemble in the form of
]1[*]1[]1[]1[*
1
]1[* )(ˆ)(ˆ )( wkykywky T
pp
q
p
,
where )...,...,( ]1*[]1*[]1*[]1*[ T
qp wwww — is a vector of adapted parameters-
synaptic weights on which additionally restrictions are set on unbiasedness
1 ]1[*]1[*
1
wIw T
qp
q
p
, (1)
where qI — )1( q is the vector of unities.
The problem of teaching the first metamodel is reduced to minimizing the
standard quadratic criterion in the presence of additional constraints (1).
Thus, the problem of training the first metamodel can be solved using the
standard method of penalty functions, which in this case reduces to minimizing
the expression
)1 ())()(())()(() , ( ]1*[2]1*[]1[]1*[]1[]1*[ wIwNYNYwNYNYwJ T
q
T , (2)
where )1())(, ),(, ),1(()( NNykyyNY T is a vector, )(1 NY
)())(ˆ, ),(ˆ, ),1(ˆ( ]1[]1[]1[ qNNykyy T is a matrix, is the penalty coefficient.
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko
ISSN 1681–6048 System Research & Information Technologies, 2024, № 1 78
Minimization (2) by ]1*[w leads to the result
q
q
T
q
LST
qLS I
INPI
wI
NPwww
)(
1
)()(lim
]1[
]1[
]1[]1[1
0
]1[*
, (3)
where ]1[LSw is a standard LSM estimate:
.)()()()()())( )(( ]1[]1[]1[]1[]1[]1[ NYNYNPNYNYNYNYw TTTLS
The same result may be obtained using the indefinite Lagrange multipliers.
Introducing into consideration meta-model error
)(ˆ)()(ˆ)()()()( ]1[ ]1*[]1*[]1*[]1[]1*[]1[ kywkyIwwkykykykyke TT
q
TT
),())(ˆ)(( ]1[]1*[]1[]1*[ kEwkykyIw T
q
T
write Lagrange function in the form
),( ]1*[wL
)1()()1()()( 1*1*1*1*1*]1[]1[]1[*
1
wIwNRwwIwkEkEw T
q
TT
q
TT
N
k
,
(where is indefinite Lagrange multiplier) and solving Kuhn–Tucker equations system
,01/),(
,0 )((2), (
]1[*]1[*
]1*[]1*[
1*
wIwL
IwNRwLV
T
q
qw
obtain the final result
.)( 2
,))(()(
1
1111*
q
T
q
q
T
qq
INRI
INRIINRw
It was proved [3; 4] that the use of score (3) leads to results that are not infe-
rior in accuracy to the best of the members of the first ensemble.
If observations from the training sample are processed sequentially online, it
is advisable to use the least squares recurrent method in the form
qpqw
IkwIIkPIkPkwkw
kykwkykykPkwkw
kykPky
kPkykykP
kPkP
p
q
LST
qq
T
q
LS
LSTLSLS
T
T
,..., 2, 1 )0(
,) 1(1()) 1()(1(1)1(
),1(ˆ)1(ˆ)1()(1()1(
,
)1(ˆ)1(ˆ1
)()1(ˆ)1(ˆ)(
)()1(
1*
111111*
111111
111
1 111
11
or if a training sample is non-stationary we may use exponentially weighted re-
current LSM method
,,..., 2, 1 )0(
,) 1(1() )(1()1()1(
),1(ˆ
)1(ˆ)()1(ˆ
)()1(ˆ)1()((
)()1(
,
)1(ˆ)()1(ˆ
)()1(ˆ)1(ˆ)(
)(
1
)1(
1*
]1[1]1[]1[]1[]1[*
]1[
]['][']['
]1[]1[]1[
1]1[
]1[]1[
]1[]1[]1[]1[
11
qpqw
IkwIIvPIkPkwkw
ky
kykPky
kwkykykP
kwkw
kykPky
kPkykykP
kPkP
p
q
LST
qq
T
q
LS
T
LST
LSLS
T
T
Hybrid system of computational intelligence based on bagging and group method …
Системні дослідження та інформаційні технології, 2024, № 1 79
where 10 — forgetting factor.
To the parameters of the metamodel can be given meaning the levels of
fuzzy membership to the optimal output signal by introducing additional restric-
tions on the non-negative values of these parameters, that is, in addition to the
configurable parameters ]1*[w we can also calculate the levels of this membership
,01 p qp ,..., 2,1 .
To do this, we introduce into consideration the extended Lagrange function
,)1())()(())()((),,( ]1[]1[]1[]1[]1[]1[]1[ TT
q
T INYNYNYNYL
where — )1( q is vector of non-negative indefinite Lagrange multipliers.
Using the equations system by Kuhn–Tucker
,0/),,(
,0),,(
]1[
]1[
1
L
LV
it’s not difficult to get the solution in the form
q
T
q
T
q
TT
q
q
T
INpI
NpINYNyNpI
INYNyNp
) (5,0
)(5,01)()(ˆ)(
),5,0 5,0)()(ˆ()(
]1[
]1[]1[
]1[]1[
For finding vector of non-negative Lagrange multipliers, it’s reasonable to
apply Arrow–Hurwitz–Uzawa procedure
)),1()()(()1(
)4(
),()1(5,0
) 1(
)()1(5,01)1(
)1()1()1(
]1[
]1[
]1[
]1[]1[
]1[]1[]1[
kkkPk
kkP
IkPI
kkPIkwI
kPkwk
r
q
T
q
T
q
LST
qLS
where (.) rP is a projector to positive ortant, )(k — learning rate, parameter.
First expression (4) after non-complex transformations may be presented in a
more compact form
)()1(5,0
) 1(
)()1(5,0
)1()1()1( ]1[
]['
]1[
]1[]1[*]1[ kkP
IkPI
kkPI
kPkwk
q
T
q
T
q
)()1(
) 1(
) 1(
5,0)1( ]1[
]1[
]1[
]1[* kkP
IkPI
IIkP
Ikw
q
T
q
T
qq
qq
where instead of least squares estimates )1(]1[ kwLS , the parameters of the met-
amodel )1(]1*[ kw are used, which simplifies the process of configuring it.
As a result of learning the first metamodel, the optimal signal )(]1*[ ky is
formed at its output, as well as q signals ) (ˆ ]1[
, ky p from which we choose
) ( 1 nqifn with the highest levels of fuzzy membership ]1[
p , which subse-
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko
ISSN 1681–6048 System Research & Information Technologies, 2024, № 1 80
quently in the form of )1( n — vector are fed to the input of the second ensem-
ble, the outputs of which go to the inputs of the second metamodel, and so on.
The process of increasing the number of ensembles and metamodels continues
until the required accuracy of the last metamodel with the output )(]*[ ky s is
achieved, or the value of the criterion minimized for the bagging model begins to
increase, i.e. ))(( ))(( ]*[ 2]1*[ 2 kyky ss .
EXPERIMENTAL INVESTIGATIONS
The experimental investigations of bagging based on GMDH were performed at
the problems of short-term and middle-term forecasting of Dow–Jones Industrial
average index. The dynamics of DJIA is presented in the Fig. 2.
The data of DJ index was taken since 05.07.22 till 03.07.23.
The correlation function of process Dow–Jones index was calculated and the
correlogram was built presented in the Fig. 3.
Analyzing this correlogram we can see strong correlation between values of DJIA.
Date
C
lo
se
Fig. 2. Dynamics of Dow–Jones Average
Lag
C
oe
ff
ic
ie
nt
Fig. 3. Correlogram
Hybrid system of computational intelligence based on bagging and group method …
Системні дослідження та інформаційні технології, 2024, № 1 81
As a first model used in bagging ensemble is ARIMA. ACF function plot is
presented in the Fig. 4.
Using differencing the process DJIA was transformed to stationary one:
Dickey–Fuller test was performed: P-value:1.3051439086544856e-28 < 0.05.
The experimental investigations were performed at different forecasting in-
tervals: 1, 3, 5 (short-term) and 10, 20 days (middle-term) forecasting.
Flow-chart for 1 day forecast is presented in the Fig. 5.
Flow chart of forecast by ARIMA for 5 days is shown in the Fig. 6.
Fig. 4. ACF plot
Fig. 6. Flow chart of forecast at 5 days interval for ARIMA
Date
V
al
ue
Fig. 5. Flow chart of forecast at 1 day interval for ARIMA
Date
V
al
ue
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko
ISSN 1681–6048 System Research & Information Technologies, 2024, № 1 82
Next model used in ensemble is Hybrid GMDH-neo-fuzzy network. It was
optimized by parameters. After that the experiments on forecasting with different
forecasting intervals 1, 3, 5, 10, 20 days were performed. Some of results are pre-
sented below. Flowchart of forecast for 3 days is shown in the Fig. 7 and for 10
days in the Fig. 8.
After that the algorithm of bagging based on GMDH was implemented
and experiments at short-term and middle-term forecasting were performed.
The tables with criteria MSE and MAPE are presented in the Tables 1 and 2.
T a b l e 1 . Average MSE values for different intervals
Interval ARIMA GMDH-neo-fuzzy HSCI-GMDH-bagging
1 17422.752 27420.394 17382.425
3 43434.022 59202.99 49332.635
5 66993.011 78330.284 58153.202
10 235427.989 108358.616 104324.0
20 696291.974 253693.345 241508.146
Fig. 8. Flow chart of forecast by GMDH-neo-fuzzy network for 10 days
Date
In
de
x
Date
In
de
x
Fig. 7. Flow chart of forecast for 3 days interval by GMDH-neo-fuzzy network
Hybrid system of computational intelligence based on bagging and group method …
Системні дослідження та інформаційні технології, 2024, № 1 83
T a b l e 2 . Average MAPE values for different intervals
Interval ARIMA GMDH-neo-fuzzy HSCI-GMDH-bagging
1 0.83 1.049 0.828
3 1.416 1.63 1.398
5 1.616 1.903 1.577
10 3.134 2.208 2.108
20 5.579 3.433 3.308
Average MSE values for different intervals are presented in the Fig. 9 and
MAPE values — in the Fig. 10.
Analyzing the presented results, one may conclude that bagging procedure
based on GMDH has the best results as compared with separate models in ensem-
ble, the second place takes GMDH-neo-fuzzy network for all the intervals at the
exception 1, 3 and 5 days. And for short-term forecasting 1, 3 and 5 days ARIMA
appears to be better than hybrid GMDH-neo-fuzzy network.
Fig. 9. Average MSE values for different forecasting intervals
Fig. 10. Average MAPE values for different forecasting intervals
Ye. Bodyanskiy, O. Kuzmenko, He. Zaichenko, Yu. Zaychenko
ISSN 1681–6048 System Research & Information Technologies, 2024, № 1 84
In a whole the obtained results well comply with theoretical statements as
for properties of bagging procedure and the proposed bagging based on GMDH
in HSCI system appeared to be very efficient procedure which demand minimum
calculations due to application of GMDH.
CONCLUSION
The architecture and learning algorithms of hybrid computational intelligence sys-
tem, which is built based on Group Method of Data Handling (GMDH) method
and bagging approach, are proposed.
The system consists of a sequence of stacks, while odd stacks are essentially
ensembles formed by parallel connected different subsystems that solve the same
problem, while even ones are metamodels that implement the bagging procedure
and calculate the levels of fuzzy membership of each member of the ensemble to
the optimal result. The process of increasing the number of stacks is based on
GMDH principles until the desired accuracy of the final results is achieved. The
proposed system does not require large volumes of training samples, provides
online work and automatically determines the number of its layers — stacks in the
learning process.
Experimental investigations confirm that Hybrid system of computational in-
telligence based on bagging and GMDH is effective for short- and middle-term
forecasting in the financial sphere and has better metrics than ARIMA and
GMDH neo fuzzy.
REFERENCES
1. G. Cybenko, “Approximation by superpositions of a sigmoidal function,”
Mathematics of Control, Signals and Systems, vol. 2, pр. 303–314, 1989.
2. L.K. Hansen, P. Salоmon, “Neural network ensembles,” IEEE Trans. on Pattern
Analysis and Machine Intelligence, vol. 12, pp. 993–1000,1990.
3. Ye. Bodyanskiy, I. Pliss, “Adaptive generalized forecasting of multivariate stochas-
tic signals,” Proc. Latvian Sign. Proc. Int. Conf. Riga, vol. 2, pp. 80–83, 1990.
4. Ye.V. Bodyanskiy, I.A. Rudneva, “On one adaptive algorithm for detecting discords in
random sequences,” Automation and Remote Control, 56, no. 10, pp. 1439–1443, 1995.
5. A.J.C. Sharkey, “On combining artificial neural nets,” Connect. Sci., vol. 8, pp. 299–313,
1996.
6. S. Hashem, “Optimal linear combination of neural networks,” Neural Networks,
vol. 10, no. 4, pp. 599–614, 1997.
7. U. Naftaly, N. Intrator, and D. Horn, “Optimal ensemble averaging of neural net-
works,” Network: Comput. Neural Syst., vol. 8, pр. 283–296, 1997.
8. D.T. Pham, X. Liu, “Neural Networks for Identification,” Prediction and Control.
London, Springer Verlag, 1995, 238 p.
9. L. Breiman, Bagging Predictors. Techn. Report No. 421. Dept. of Statistics, Univ. of
California, Berkeley, CA 94720, 1994, 19 р.
10. A.G. Ivakhnenko, V.G. Lapa, Cybernetic forecasting devices. K.: “Naukova
Dumka”, 1965, 216 p.
11. A.G. Ivakhnenko, G.A. Ivakhnenko, and J.A. Mueller, “Self-organization of the neu-
ral networks with active neurons,” Pattern Recognition and Image Analysis, 4(2),
pp. 177–188, 1994.
12. J. Schmidhuber, “Deep learning in neural networks: An overview,” Neural Net-
works, vol. 61, pp. 85–117, 2015.
Hybrid system of computational intelligence based on bagging and group method …
Системні дослідження та інформаційні технології, 2024, № 1 85
13. Yuriy Zaychenko, Yevgeniy Bodyanskiy, Oleksii Tyshchenko, Olena Boiko, and
Galib Hamidov, “Hybrid GMDH-neuro-fuzzy system and its training scheme,” Int.
Journal Information Theories and Applications, vol. 24, no. 2, pp. 156–172, 2018.
14. Yuriy Zaychenko, Yevgeniy Bodyanskiy, Olena Boiko, and Galib Hamidov, “Evolv-
ing Hybrid GMDH-NeuroFuzzy Network and Its Application,” International Con-
ference IEEE-SAIC 2018, Kyiv, IASA, 8–11 October, 2018.
15. Yevgeniy Bodyanskiy, Nonna Kulishova, Yuriy Zaychenko, and Galib Hamidov,
“Spline-Orthogonal Extended Neo-Fuzzy Neuron,” International conference CISP-
BMEI 2019.
16. Yevgeniy Bodyanskiy, Yuriy Zaychenko, Olena Boiko, Galib Hamidov, and Anna
Zelikman, “The Hybrid GMDH-Neo-fuzzy Neural Network in Forecasting Problems
in Financial Sphere,” Intern. Conference IEEE SAIC 2020 in book “Advances in
Intelligent Computing”, Springer, 2020, vol. 1075, pp. 221–225.
17. Ye. Bodyanskiy, O. Vynokurova, and I. Pliss, “Hybrid GMDH-neural network of
computational intelligence,” Proc. 3rd Int. Workshop on Inductive Modeling,
Krynica, Poland, 2009, pp. 100–107.
Received 15.08.2023
INFORMATION ON THE ARTICLE
Yevgeniy V. Bodyanskiy, ORCID: 0000-0001-5418-2143, Kharkiv National University
of Radio Electronics, Ukraine, e-mail: yevgeniy.bodyanskiy@nure.ua
Oleksii V. Kuzmenko, ORCID: 0000-0003-1581-6224, Educational and Research Insti-
tute for Applied System Analysis of the National Technical University of Ukraine “Igor
Sikorsky Kyiv Polytechnic Institute”, Ukraine, e-mail: oleksii.kuzmenko@ukr.net
Helen Yu. Zaichenko, ORCID: 0000-0002-4630-5155, Educational and Research Insti-
tute for Applied System Analysis of the National Technical University of Ukraine “Igor
Sikorsky Kyiv Polytechnic Institute”, Ukraine, e-mail: syncmaster@bigmir.net
Yuriy P. Zaychenko, ORCID: 0000-0001-9662-3269, Educational and Research Institute
for Applied System Analysis of the National Technical University of Ukraine “Igor Sikor-
sky Kyiv Polytechnic Institute”, Ukraine, e-mail: zaychenkoyuri@ukr.net
ГІБРИДНА СИСТЕМА ОБЧИСЛЮВАЛЬНОГО ІНТЕЛЕКТУ НА ОСНОВІ
БЕГГІНГУ ТА МЕТОДУ ГРУПОВОГО УРАХУВАННЯ АРГУМЕНТІВ /
Є.В.Бодянський, О.В. Кузьменко, Ю.П. Зайченко, О.Ю. Зайченко
Анотація. Розглянуто проблему короткострокового та середньострокового
прогнозування у фінансовій сфері. Для її вирішення запропоновано гібридну
систему обчислювального інтелекту на основі методу групового урахування
аргументів (МГУА) та беггінгу, а також алгоритм її навчання. Непарні стеки
гібридної системи сформовані ансамблями паралельно з’єднаних підсистем.
Як такі підсистеми обрано ARIMA та гібридну мережу МГУА-нео-фаззі. За-
пропонована система не потребує великого обсягу навчальної вибірки, автома-
тично визначає кількість стеків у процесі навчання та забезпечує роботу у ре-
жимі online. Проведено експериментальні дослідження з використанням
запропонованої гібридної системи, а також окремо ARIMA та МГУА-нео-
фаззі. Порівняно точність прогнозів, отриманих експериментальним шля-
хом, на основі чого обґрунтовано доцільність застосування запропонованої
гібридної системи.
Ключові слова: гібридна система, беггінг, гібридна мережа МГУА-нео-фаззі,
ARIMA, короткострокове та середньострокове прогнозування.
|
| id | journaliasakpiua-article-304428 |
| institution | System research and information technologies |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2025-07-17T10:28:30Z |
| publishDate | 2024 |
| publisher | The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" |
| record_format | ojs |
| resource_txt_mv | journaliasakpiua/d3/3a69fc2b528d69303220760a17652ed3.pdf |
| spelling | journaliasakpiua-article-3044282024-05-23T07:09:36Z Hybrid system of computational intelligence based on bagging and group method of data handling Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів Bodyanskiy, Yevgeniy Kuzmenko, Oleksii Zaichenko, Helen Zaychenko, Yuriy гібридна система беггінг гібридна мережа МГУА-нео-фаззі ARIMA короткострокове та середньострокове прогнозування hybrid system bagging hybrid GMDH-neo-fuzzy network ARIMA short- and middle-term forecasting The paper considers the problem of short- and middle-term forecasting in the financial sphere. To solve this problem, a hybrid system of computational intelligence based on the group method of data handling (GMDH) and bagging, as well as an algorithm for its training, is proposed. The odd stacks of the hybrid system are formed by ensembles of parallel connected subsystems. ARIMA and the GMDH-neo-fuzzy hybrid network were chosen as such subsystems. The proposed system does not require a large training data set, automatically determines the number of stacks during training, and provides online operation. The experimental investigations were conducted using the proposed hybrid system, as well as separately using ARIMA and GMDH-neo-fuzzy. The accuracy of the predictions obtained is compared, based on which the feasibility of using the proposed hybrid system is substantiated. Розглянуто проблему короткострокового та середньострокового прогнозування у фінансовій сфері. Для її вирішення запропоновано гібридну систему обчислювального інтелекту на основі методу групового урахування аргументів (МГУА) та беггінгу, а також алгоритм її навчання. Непарні стеки гібридної системи сформовані ансамблями паралельно з’єднаних підсистем. Як такі підсистеми обрано ARIMA та гібридну мережу МГУА-нео-фаззі. Запропонована система не потребує великого обсягу навчальної вибірки, автоматично визначає кількість стеків у процесі навчання та забезпечує роботу у режимі online. Проведено експериментальні дослідження з використанням запропонованої гібридної системи, а також окремо ARIMA та МГУА-нео-фаззі. Порівняно точність прогнозів, отриманих експериментальним шляхом, на основі чого обґрунтовано доцільність застосування запропонованої гібридної системи. The National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" 2024-03-29 Article Article application/pdf https://journal.iasa.kpi.ua/article/view/304428 10.20535/SRIT.2308-8893.2024.1.06 System research and information technologies; No. 1 (2024); 75-85 Системные исследования и информационные технологии; № 1 (2024); 75-85 Системні дослідження та інформаційні технології; № 1 (2024); 75-85 2308-8893 1681-6048 en https://journal.iasa.kpi.ua/article/view/304428/296331 |
| spellingShingle | гібридна система беггінг гібридна мережа МГУА-нео-фаззі ARIMA короткострокове та середньострокове прогнозування Bodyanskiy, Yevgeniy Kuzmenko, Oleksii Zaichenko, Helen Zaychenko, Yuriy Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title | Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title_alt | Hybrid system of computational intelligence based on bagging and group method of data handling |
| title_full | Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title_fullStr | Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title_full_unstemmed | Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title_short | Гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| title_sort | гібридна система обчислювального інтелекту на основі беггінгу та методу групового урахування аргументів |
| topic | гібридна система беггінг гібридна мережа МГУА-нео-фаззі ARIMA короткострокове та середньострокове прогнозування |
| topic_facet | гібридна система беггінг гібридна мережа МГУА-нео-фаззі ARIMA короткострокове та середньострокове прогнозування hybrid system bagging hybrid GMDH-neo-fuzzy network ARIMA short- and middle-term forecasting |
| url | https://journal.iasa.kpi.ua/article/view/304428 |
| work_keys_str_mv | AT bodyanskiyyevgeniy hybridsystemofcomputationalintelligencebasedonbaggingandgroupmethodofdatahandling AT kuzmenkooleksii hybridsystemofcomputationalintelligencebasedonbaggingandgroupmethodofdatahandling AT zaichenkohelen hybridsystemofcomputationalintelligencebasedonbaggingandgroupmethodofdatahandling AT zaychenkoyuriy hybridsystemofcomputationalintelligencebasedonbaggingandgroupmethodofdatahandling AT bodyanskiyyevgeniy gíbridnasistemaobčislûvalʹnogoíntelektunaosnovíbeggíngutametodugrupovogourahuvannâargumentív AT kuzmenkooleksii gíbridnasistemaobčislûvalʹnogoíntelektunaosnovíbeggíngutametodugrupovogourahuvannâargumentív AT zaichenkohelen gíbridnasistemaobčislûvalʹnogoíntelektunaosnovíbeggíngutametodugrupovogourahuvannâargumentív AT zaychenkoyuriy gíbridnasistemaobčislûvalʹnogoíntelektunaosnovíbeggíngutametodugrupovogourahuvannâargumentív |