Employing GA-based Adaptive Grey Model for learning with short-term sequence data

Der-Chiang Li, Wu Kuo Lin

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Product lifecycles are becoming shorter and shorter because of the global competition. In such highly competitive condition, the importance for enterprises to build forecasting models with short-term time series data significantly arises in recent decade. Unfortunately, since the information the short-term data provides is limited, it is a great challenge to learn its pattern behavior. To overcome the issue, the Grey Model (GM) reveals a clear direction by accumulating data values to make them represent as a monotonically increasing sequence. Accordingly, improving the forecasting accuracy of GM has remarkable developments in theories in decades. Among them, the Adaptive Grey Model (AGM) tries to find the suitable background values by means of taking the data occurring trend into consideration. In fact, setting the suitable background values of GM can be regarded as a process of searching the optimal solutions. This paper thus employs the genetic algorithm (GA) to facilitate the improvement of searching process, where the initial solutions are obtained from AGM, to build a more accurate model named as GAAGM(1.1). The experiment results show that GAAGM outperforms AGM and some other improved GM models in prediction accuracy; and furthermore, this demonstrates that searching for more suitable coefficient sets can indeed help enhance the theory structure of GM models.

Original languageEnglish
Pages (from-to)96-106
Number of pages11
JournalJournal of Grey System
Volume25
Issue number4
Publication statusPublished - 2013 Dec 1

Fingerprint

Grey Model
Genetic algorithms
Genetic Algorithm
Forecasting
Monotonic increasing sequence
Learning
Genetic algorithm
Grey model
Time Series Data
Model
Life Cycle
Optimal Solution
Time series
Prediction
Coefficient

All Science Journal Classification (ASJC) codes

  • Modelling and Simulation
  • Statistics, Probability and Uncertainty
  • Control and Optimization
  • Applied Mathematics

Cite this

@article{3eaacaf5f3a64801944c6e69aa2cf0f0,
title = "Employing GA-based Adaptive Grey Model for learning with short-term sequence data",
abstract = "Product lifecycles are becoming shorter and shorter because of the global competition. In such highly competitive condition, the importance for enterprises to build forecasting models with short-term time series data significantly arises in recent decade. Unfortunately, since the information the short-term data provides is limited, it is a great challenge to learn its pattern behavior. To overcome the issue, the Grey Model (GM) reveals a clear direction by accumulating data values to make them represent as a monotonically increasing sequence. Accordingly, improving the forecasting accuracy of GM has remarkable developments in theories in decades. Among them, the Adaptive Grey Model (AGM) tries to find the suitable background values by means of taking the data occurring trend into consideration. In fact, setting the suitable background values of GM can be regarded as a process of searching the optimal solutions. This paper thus employs the genetic algorithm (GA) to facilitate the improvement of searching process, where the initial solutions are obtained from AGM, to build a more accurate model named as GAAGM(1.1). The experiment results show that GAAGM outperforms AGM and some other improved GM models in prediction accuracy; and furthermore, this demonstrates that searching for more suitable coefficient sets can indeed help enhance the theory structure of GM models.",
author = "Der-Chiang Li and Lin, {Wu Kuo}",
year = "2013",
month = "12",
day = "1",
language = "English",
volume = "25",
pages = "96--106",
journal = "Journal of Grey System",
issn = "0957-3720",
publisher = "Research Information Ltd",
number = "4",

}

Employing GA-based Adaptive Grey Model for learning with short-term sequence data. / Li, Der-Chiang; Lin, Wu Kuo.

In: Journal of Grey System, Vol. 25, No. 4, 01.12.2013, p. 96-106.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Employing GA-based Adaptive Grey Model for learning with short-term sequence data

AU - Li, Der-Chiang

AU - Lin, Wu Kuo

PY - 2013/12/1

Y1 - 2013/12/1

N2 - Product lifecycles are becoming shorter and shorter because of the global competition. In such highly competitive condition, the importance for enterprises to build forecasting models with short-term time series data significantly arises in recent decade. Unfortunately, since the information the short-term data provides is limited, it is a great challenge to learn its pattern behavior. To overcome the issue, the Grey Model (GM) reveals a clear direction by accumulating data values to make them represent as a monotonically increasing sequence. Accordingly, improving the forecasting accuracy of GM has remarkable developments in theories in decades. Among them, the Adaptive Grey Model (AGM) tries to find the suitable background values by means of taking the data occurring trend into consideration. In fact, setting the suitable background values of GM can be regarded as a process of searching the optimal solutions. This paper thus employs the genetic algorithm (GA) to facilitate the improvement of searching process, where the initial solutions are obtained from AGM, to build a more accurate model named as GAAGM(1.1). The experiment results show that GAAGM outperforms AGM and some other improved GM models in prediction accuracy; and furthermore, this demonstrates that searching for more suitable coefficient sets can indeed help enhance the theory structure of GM models.

AB - Product lifecycles are becoming shorter and shorter because of the global competition. In such highly competitive condition, the importance for enterprises to build forecasting models with short-term time series data significantly arises in recent decade. Unfortunately, since the information the short-term data provides is limited, it is a great challenge to learn its pattern behavior. To overcome the issue, the Grey Model (GM) reveals a clear direction by accumulating data values to make them represent as a monotonically increasing sequence. Accordingly, improving the forecasting accuracy of GM has remarkable developments in theories in decades. Among them, the Adaptive Grey Model (AGM) tries to find the suitable background values by means of taking the data occurring trend into consideration. In fact, setting the suitable background values of GM can be regarded as a process of searching the optimal solutions. This paper thus employs the genetic algorithm (GA) to facilitate the improvement of searching process, where the initial solutions are obtained from AGM, to build a more accurate model named as GAAGM(1.1). The experiment results show that GAAGM outperforms AGM and some other improved GM models in prediction accuracy; and furthermore, this demonstrates that searching for more suitable coefficient sets can indeed help enhance the theory structure of GM models.

UR - http://www.scopus.com/inward/record.url?scp=84893578207&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84893578207&partnerID=8YFLogxK

M3 - Article

VL - 25

SP - 96

EP - 106

JO - Journal of Grey System

JF - Journal of Grey System

SN - 0957-3720

IS - 4

ER -