A Weighted Fuzzy Time Series Forecasting Model

Daniel Ortiz-Arroyo, Jens Runi Poulsen

Research output: Contribution to journalConference article in JournalResearchpeer-review

103 Downloads (Pure)

Abstract

In this paper we describe a new automatic partitioning method and a first order weighted fuzzy time series forecasting model. First, we show that our automatic fuzzy partitioning method provides an accurate approximation to the original time series. The fuzzy sets extracted from our partitioning are grouped to create a rule-base that will be used in forecasting. We found that the accuracy of our first order model is improved when an ordered weighting averaging operator is applied. The model presented in this paper does not attempt to produce the most accurate forecasting results, when compared with other more complex higher order models. Our goal is to show that there is still space for improvement when simple first order forecasting models are used. Our results show that the combination of a simple partitioning method, a first order model, and an averaging operator is still capable of outperforming not only the first order models that have proposed in the literature but also other higher order models.
Original languageEnglish
JournalIndian Journal of Science and Technology
Volume11
Issue number27
Pages (from-to)1-11
Number of pages11
ISSN0974-6846
DOIs
Publication statusPublished - Jul 2018

Fingerprint

Time series
Fuzzy sets

Keywords

  • Time Series Analysis
  • Fuzzy logic
  • Aggregation Operators

Cite this

@inproceedings{fa7d04bb69b7492dab33900667ea8b16,
title = "A Weighted Fuzzy Time Series Forecasting Model",
abstract = "In this paper we describe a new automatic partitioning method and a first order weighted fuzzy time series forecasting model. First, we show that our automatic fuzzy partitioning method provides an accurate approximation to the original time series. The fuzzy sets extracted from our partitioning are grouped to create a rule-base that will be used in forecasting. We found that the accuracy of our first order model is improved when an ordered weighting averaging operator is applied. The model presented in this paper does not attempt to produce the most accurate forecasting results, when compared with other more complex higher order models. Our goal is to show that there is still space for improvement when simple first order forecasting models are used. Our results show that the combination of a simple partitioning method, a first order model, and an averaging operator is still capable of outperforming not only the first order models that have proposed in the literature but also other higher order models.",
keywords = "Time Series Analysis, Fuzzy logic, Aggregation Operators",
author = "Daniel Ortiz-Arroyo and Poulsen, {Jens Runi}",
year = "2018",
month = "7",
doi = "10.17485/ijst/2018/v11i27/130708",
language = "English",
volume = "11",
pages = "1--11",
journal = "Indian Journal of Science and Technology",
issn = "0974-6846",
publisher = "Indian Society for Education and Environment",
number = "27",

}

A Weighted Fuzzy Time Series Forecasting Model. / Ortiz-Arroyo, Daniel; Poulsen, Jens Runi.

In: Indian Journal of Science and Technology, Vol. 11, No. 27, 07.2018, p. 1-11.

Research output: Contribution to journalConference article in JournalResearchpeer-review

TY - GEN

T1 - A Weighted Fuzzy Time Series Forecasting Model

AU - Ortiz-Arroyo, Daniel

AU - Poulsen, Jens Runi

PY - 2018/7

Y1 - 2018/7

N2 - In this paper we describe a new automatic partitioning method and a first order weighted fuzzy time series forecasting model. First, we show that our automatic fuzzy partitioning method provides an accurate approximation to the original time series. The fuzzy sets extracted from our partitioning are grouped to create a rule-base that will be used in forecasting. We found that the accuracy of our first order model is improved when an ordered weighting averaging operator is applied. The model presented in this paper does not attempt to produce the most accurate forecasting results, when compared with other more complex higher order models. Our goal is to show that there is still space for improvement when simple first order forecasting models are used. Our results show that the combination of a simple partitioning method, a first order model, and an averaging operator is still capable of outperforming not only the first order models that have proposed in the literature but also other higher order models.

AB - In this paper we describe a new automatic partitioning method and a first order weighted fuzzy time series forecasting model. First, we show that our automatic fuzzy partitioning method provides an accurate approximation to the original time series. The fuzzy sets extracted from our partitioning are grouped to create a rule-base that will be used in forecasting. We found that the accuracy of our first order model is improved when an ordered weighting averaging operator is applied. The model presented in this paper does not attempt to produce the most accurate forecasting results, when compared with other more complex higher order models. Our goal is to show that there is still space for improvement when simple first order forecasting models are used. Our results show that the combination of a simple partitioning method, a first order model, and an averaging operator is still capable of outperforming not only the first order models that have proposed in the literature but also other higher order models.

KW - Time Series Analysis

KW - Fuzzy logic

KW - Aggregation Operators

U2 - 10.17485/ijst/2018/v11i27/130708

DO - 10.17485/ijst/2018/v11i27/130708

M3 - Conference article in Journal

VL - 11

SP - 1

EP - 11

JO - Indian Journal of Science and Technology

JF - Indian Journal of Science and Technology

SN - 0974-6846

IS - 27

ER -