Extrapolation-based grey model for small-data-set forecasting

Che Jung Chang, Der Chiang Li, Chien Chih Chen, Wen Chih Chen

Research output: Contribution to journalArticle

Abstract

Product life cycles have become increasingly shorter owing to the rise of global competition in recent decades. Competitive tension is especially high in electronics-related industries. It is usually difficult for most enterprises to collect sufficient quantities of samples with which to obtain useful information when making decisions in such a highly competitive environment. Grey system theory plays a vital role in addressing the issue of insufficient sample quantities. The traditional GM(1,1) model is well known for its ability to generate useful forecasts with a small quantity of samples; however, the newest datum is always weakened to alleviate the randomness of data in the traditional GM(1,1) model, causing it to output higher prediction errors. To overcome such imperfections, this study proposes a modified grey forecasting model named EP-GM(1,1), in which a new equation for calculating the background values in the traditional GM(1,1) model is developed based on linear extrapolation to emphasize the importance of the newest datum. To evaluate the forecasting ability of EP-GM(1,1), the monthly demand of thin-film-transistor liquid-crystal display panels were employed for experimentation. The results indicate that EP-GM(1,1) can engender a favorable prediction result, demonstrating that the model is a feasible tool for small-sample forecasting.

Original languageEnglish
Pages (from-to)171-182
Number of pages12
JournalEconomic Computation and Economic Cybernetics Studies and Research
Volume53
Issue number1
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

GM(1,1)
GM(1,1) Model
Grey Model
Extrapolation
Forecasting
Grey System Theory
Thin-film Transistor
Liquid Crystal Display
Imperfections
Prediction Error
Small Sample
Life Cycle
Experimentation
Randomness
Forecast
Decision Making
Electronics
Industry
Sufficient
System theory

All Science Journal Classification (ASJC) codes

  • Economics and Econometrics
  • Computer Science Applications
  • Applied Mathematics

Cite this

@article{d430db1259074da9add9cc7ddcd74c7a,
title = "Extrapolation-based grey model for small-data-set forecasting",
abstract = "Product life cycles have become increasingly shorter owing to the rise of global competition in recent decades. Competitive tension is especially high in electronics-related industries. It is usually difficult for most enterprises to collect sufficient quantities of samples with which to obtain useful information when making decisions in such a highly competitive environment. Grey system theory plays a vital role in addressing the issue of insufficient sample quantities. The traditional GM(1,1) model is well known for its ability to generate useful forecasts with a small quantity of samples; however, the newest datum is always weakened to alleviate the randomness of data in the traditional GM(1,1) model, causing it to output higher prediction errors. To overcome such imperfections, this study proposes a modified grey forecasting model named EP-GM(1,1), in which a new equation for calculating the background values in the traditional GM(1,1) model is developed based on linear extrapolation to emphasize the importance of the newest datum. To evaluate the forecasting ability of EP-GM(1,1), the monthly demand of thin-film-transistor liquid-crystal display panels were employed for experimentation. The results indicate that EP-GM(1,1) can engender a favorable prediction result, demonstrating that the model is a feasible tool for small-sample forecasting.",
author = "Chang, {Che Jung} and Li, {Der Chiang} and Chen, {Chien Chih} and Chen, {Wen Chih}",
year = "2019",
month = "1",
day = "1",
doi = "10.24818/18423264/53.1.19.11",
language = "English",
volume = "53",
pages = "171--182",
journal = "Economic Computation and Economic Cybernetics Studies and Research",
issn = "0585-7511",
publisher = "Editura Academia de studii economice",
number = "1",

}

Extrapolation-based grey model for small-data-set forecasting. / Chang, Che Jung; Li, Der Chiang; Chen, Chien Chih; Chen, Wen Chih.

In: Economic Computation and Economic Cybernetics Studies and Research, Vol. 53, No. 1, 01.01.2019, p. 171-182.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Extrapolation-based grey model for small-data-set forecasting

AU - Chang, Che Jung

AU - Li, Der Chiang

AU - Chen, Chien Chih

AU - Chen, Wen Chih

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Product life cycles have become increasingly shorter owing to the rise of global competition in recent decades. Competitive tension is especially high in electronics-related industries. It is usually difficult for most enterprises to collect sufficient quantities of samples with which to obtain useful information when making decisions in such a highly competitive environment. Grey system theory plays a vital role in addressing the issue of insufficient sample quantities. The traditional GM(1,1) model is well known for its ability to generate useful forecasts with a small quantity of samples; however, the newest datum is always weakened to alleviate the randomness of data in the traditional GM(1,1) model, causing it to output higher prediction errors. To overcome such imperfections, this study proposes a modified grey forecasting model named EP-GM(1,1), in which a new equation for calculating the background values in the traditional GM(1,1) model is developed based on linear extrapolation to emphasize the importance of the newest datum. To evaluate the forecasting ability of EP-GM(1,1), the monthly demand of thin-film-transistor liquid-crystal display panels were employed for experimentation. The results indicate that EP-GM(1,1) can engender a favorable prediction result, demonstrating that the model is a feasible tool for small-sample forecasting.

AB - Product life cycles have become increasingly shorter owing to the rise of global competition in recent decades. Competitive tension is especially high in electronics-related industries. It is usually difficult for most enterprises to collect sufficient quantities of samples with which to obtain useful information when making decisions in such a highly competitive environment. Grey system theory plays a vital role in addressing the issue of insufficient sample quantities. The traditional GM(1,1) model is well known for its ability to generate useful forecasts with a small quantity of samples; however, the newest datum is always weakened to alleviate the randomness of data in the traditional GM(1,1) model, causing it to output higher prediction errors. To overcome such imperfections, this study proposes a modified grey forecasting model named EP-GM(1,1), in which a new equation for calculating the background values in the traditional GM(1,1) model is developed based on linear extrapolation to emphasize the importance of the newest datum. To evaluate the forecasting ability of EP-GM(1,1), the monthly demand of thin-film-transistor liquid-crystal display panels were employed for experimentation. The results indicate that EP-GM(1,1) can engender a favorable prediction result, demonstrating that the model is a feasible tool for small-sample forecasting.

UR - http://www.scopus.com/inward/record.url?scp=85066755116&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85066755116&partnerID=8YFLogxK

U2 - 10.24818/18423264/53.1.19.11

DO - 10.24818/18423264/53.1.19.11

M3 - Article

AN - SCOPUS:85066755116

VL - 53

SP - 171

EP - 182

JO - Economic Computation and Economic Cybernetics Studies and Research

JF - Economic Computation and Economic Cybernetics Studies and Research

SN - 0585-7511

IS - 1

ER -