Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction

Ajla Kulaglic; B. Berk Ustundag

doi:10.18421/TEM104-61

Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction

Ajla Kulaglic^*, B. Berk Ustundag

^*Corresponding author for this work

Department of Computer Engineering

Istanbul Technical University

Research output: Contribution to journal › Article › peer-review

5 Citations (Scopus)

Abstract

Multivariable machine learning (ML) models are increasingly used for time series predictions. However, avoiding the overfitting and underfitting in ML-based time series prediction requires special consideration depending on the size and characteristics of the available training dataset. Predictive error compensating wavelet neural network (PEC-WNN) improves the time series prediction accuracy by enhancing the orthogonal features within a data fusion scheme. In this study, time series prediction performance of the PEC-WNNs have been evaluated on two different problems in comparison to conventional machine learning methods including the long short-term memory (LSTM) network. The results have shown that PECNET provides significantly more accurate predictions. RMSPE error is reduced by more than 60% with respect to other compared ML methods for Lorenz Attractor and wind speed prediction problems.

Original language	English
Pages (from-to)	1955-1963
Number of pages	9
Journal	TEM Journal
Volume	10
Issue number	4
DOIs	https://doi.org/10.18421/TEM104-61
Publication status	Published - 2021

Bibliographical note

Publisher Copyright:
© 2021. TEM Journal. All Rights Reserved.

Funding

This work was supported by the research project ‘‘Development of Cyberdroid Based on Cognitive Intelligent System Applications’’ (2019-2020) that is funded by Crypttech Inc within the contract by ITUNOVA-Istanbul Technical University Technology Transfer Office. Meteorological data has been provided from Agricultural Monitoring and Information Systems Project (TARBIL 2011A020100) supported by the Turkish Ministry of Development in cooperation with Ministry of Agriculture and Forestry in Turkey. This work was supported by the research project ?Development of Cyberdroid Based on Cognitive Intelligent System Applications?? (2019-2020) that is funded by Crypttech Inc within the contract by ITUNOVA Istanbul Technical University Technology Transfer Office. Meteorological data has been provided from Agricultural Monitoring and Information Systems Project (TARBIL 2011A020100) supported by the Turkish Ministry of Development in cooperation with Ministry of Agriculture and Forestry in Turkey.

Funders	Funder number
Crypttech Inc
ITUNOVA Istanbul Technical University Technology Transfer Office
ITUNOVA-Istanbul Technical University	2011A020100
Ministry of Agriculture and Forestry in Turkey

Keywords

Multivariable time series prediction
Neural networks
Predictive error compensated wavelet neural networks
Spatial dimension
Time series prediction
Wavelet transform

Access to Document

10.18421/TEM104-61

Cite this

@article{2098ce4896a74ca0847a14d10bb6e8ba,

title = "Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction",

abstract = "Multivariable machine learning (ML) models are increasingly used for time series predictions. However, avoiding the overfitting and underfitting in ML-based time series prediction requires special consideration depending on the size and characteristics of the available training dataset. Predictive error compensating wavelet neural network (PEC-WNN) improves the time series prediction accuracy by enhancing the orthogonal features within a data fusion scheme. In this study, time series prediction performance of the PEC-WNNs have been evaluated on two different problems in comparison to conventional machine learning methods including the long short-term memory (LSTM) network. The results have shown that PECNET provides significantly more accurate predictions. RMSPE error is reduced by more than 60% with respect to other compared ML methods for Lorenz Attractor and wind speed prediction problems.",

keywords = "Multivariable time series prediction, Neural networks, Predictive error compensated wavelet neural networks, Spatial dimension, Time series prediction, Wavelet transform",

author = "Ajla Kulaglic and Ustundag, {B. Berk}",

year = "2021",

doi = "10.18421/TEM104-61",

language = "English",

volume = "10",

pages = "1955--1963",

journal = "TEM Journal",

issn = "2217-8309",

publisher = "UIKTEN - Association for Information Communication Technology Education and Science",

number = "4",

}

TY - JOUR

T1 - Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction

AU - Kulaglic, Ajla

AU - Ustundag, B. Berk

PY - 2021

Y1 - 2021

N2 - Multivariable machine learning (ML) models are increasingly used for time series predictions. However, avoiding the overfitting and underfitting in ML-based time series prediction requires special consideration depending on the size and characteristics of the available training dataset. Predictive error compensating wavelet neural network (PEC-WNN) improves the time series prediction accuracy by enhancing the orthogonal features within a data fusion scheme. In this study, time series prediction performance of the PEC-WNNs have been evaluated on two different problems in comparison to conventional machine learning methods including the long short-term memory (LSTM) network. The results have shown that PECNET provides significantly more accurate predictions. RMSPE error is reduced by more than 60% with respect to other compared ML methods for Lorenz Attractor and wind speed prediction problems.

AB - Multivariable machine learning (ML) models are increasingly used for time series predictions. However, avoiding the overfitting and underfitting in ML-based time series prediction requires special consideration depending on the size and characteristics of the available training dataset. Predictive error compensating wavelet neural network (PEC-WNN) improves the time series prediction accuracy by enhancing the orthogonal features within a data fusion scheme. In this study, time series prediction performance of the PEC-WNNs have been evaluated on two different problems in comparison to conventional machine learning methods including the long short-term memory (LSTM) network. The results have shown that PECNET provides significantly more accurate predictions. RMSPE error is reduced by more than 60% with respect to other compared ML methods for Lorenz Attractor and wind speed prediction problems.

KW - Multivariable time series prediction

KW - Neural networks

KW - Predictive error compensated wavelet neural networks

KW - Spatial dimension

KW - Time series prediction

KW - Wavelet transform

UR - http://www.scopus.com/inward/record.url?scp=85121766001&partnerID=8YFLogxK

U2 - 10.18421/TEM104-61

DO - 10.18421/TEM104-61

M3 - Article

AN - SCOPUS:85121766001

SN - 2217-8309

VL - 10

SP - 1955

EP - 1963

JO - TEM Journal

JF - TEM Journal

IS - 4

ER -

Predictive Error Compensating Wavelet Neural Network Model for Multivariable Time Series Prediction

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this