A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization

Abdul K. Mohamed; Fulufhelo V. Nelwamondo; Tshilidzi Marwala

A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization

Abdul K. Mohamed, Fulufhelo V. Nelwamondo, Tshilidzi Marwala

University of the Witwatersrand

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

This paper presents a hybrid auto-associative neural networks that combines the functionality of radial basis functions, multilayer perceprons and principal component analysis. This hybrid network is developed and its performance in conjunction with the GA is compared to that of an ordinary AANN. A PCA and neural network missing data estimation system is also developed and compared to the other two systems. A description of the methodology is presented followed by the experimental implementation using HIV data from the department of Health. Accuracies of up to 95% are achieved on imputation of some of the variables.

Original language	English
Title of host publication	WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.
Pages	30-35
Number of pages	6
Publication status	Published - 2008
Externally published	Yes
Event	12th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2008, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Orlando, FL, United States Duration: 29 Jun 2008 → 2 Jul 2008

Publication series

Name	WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.
Volume	5

Conference

Conference	12th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2008, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008
Country/Territory	United States
City	Orlando, FL
Period	29/06/08 → 2/07/08

ASJC Scopus subject areas

Artificial Intelligence
Computer Networks and Communications

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Cite this

Mohamed, A. K., Nelwamondo, F. V., & Marwala, T. (2008). A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization. In WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc. (pp. 30-35). (WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.; Vol. 5).

Mohamed, Abdul K. ; Nelwamondo, Fulufhelo V. ; Marwala, Tshilidzi. / A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization. WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.. 2008. pp. 30-35 (WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.).

@inproceedings{0d18f256dd5e4de2b0df9f45c8c0cb74,

title = "A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization",

abstract = "This paper presents a hybrid auto-associative neural networks that combines the functionality of radial basis functions, multilayer perceprons and principal component analysis. This hybrid network is developed and its performance in conjunction with the GA is compared to that of an ordinary AANN. A PCA and neural network missing data estimation system is also developed and compared to the other two systems. A description of the methodology is presented followed by the experimental implementation using HIV data from the department of Health. Accuracies of up to 95% are achieved on imputation of some of the variables.",

author = "Mohamed, {Abdul K.} and Nelwamondo, {Fulufhelo V.} and Tshilidzi Marwala",

year = "2008",

language = "English",

isbn = "1934272353",

series = "WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.",

pages = "30--35",

booktitle = "WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.",

note = "12th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2008, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 ; Conference date: 29-06-2008 Through 02-07-2008",

}

Mohamed, AK, Nelwamondo, FV & Marwala, T 2008, A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization. in WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.. WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc., vol. 5, pp. 30-35, 12th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2008, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008, Orlando, FL, United States, 29/06/08.

A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization. / Mohamed, Abdul K.; Nelwamondo, Fulufhelo V.; Marwala, Tshilidzi.
WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.. 2008. p. 30-35 (WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.; Vol. 5).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization

AU - Mohamed, Abdul K.

AU - Nelwamondo, Fulufhelo V.

AU - Marwala, Tshilidzi

PY - 2008

Y1 - 2008

N2 - This paper presents a hybrid auto-associative neural networks that combines the functionality of radial basis functions, multilayer perceprons and principal component analysis. This hybrid network is developed and its performance in conjunction with the GA is compared to that of an ordinary AANN. A PCA and neural network missing data estimation system is also developed and compared to the other two systems. A description of the methodology is presented followed by the experimental implementation using HIV data from the department of Health. Accuracies of up to 95% are achieved on imputation of some of the variables.

AB - This paper presents a hybrid auto-associative neural networks that combines the functionality of radial basis functions, multilayer perceprons and principal component analysis. This hybrid network is developed and its performance in conjunction with the GA is compared to that of an ordinary AANN. A PCA and neural network missing data estimation system is also developed and compared to the other two systems. A description of the methodology is presented followed by the experimental implementation using HIV data from the department of Health. Accuracies of up to 95% are achieved on imputation of some of the variables.

UR - http://www.scopus.com/inward/record.url?scp=84869409393&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84869409393

SN - 1934272353

SN - 9781934272350

T3 - WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.

SP - 30

EP - 35

BT - WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.

T2 - 12th World Multi-Conference on Systemics, Cybernetics and Informatics, WMSCI 2008, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008

Y2 - 29 June 2008 through 2 July 2008

ER -

Mohamed AK, Nelwamondo FV, Marwala T. A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization. In WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.. 2008. p. 30-35. (WMSCI 2008 - The 12th World Multi-Conference on Systemics, Cybernetics and Informatics, Jointly with the 14th International Conference on Information Systems Analysis and Synthesis, ISAS 2008 - Proc.).

A hybrid approach to estimation of missing data through the use of neural networks, principal component analysis and stochastic optimization

Abstract

Publication series

Conference

ASJC Scopus subject areas

UN SDGs

Other files and links

Fingerprint

Cite this