A low cost and highly accurate technique for big data spatial-temporal interpolation

M. Esmaeilbeigi; O. Chatrabgoun; A. Hosseinian-Far; R. Montasari; A. Daneshkhah

doi:10.1016/j.apnum.2020.03.009

A low cost and highly accurate technique for big data spatial-temporal interpolation

M. Esmaeilbeigi, O. Chatrabgoun, A. Hosseinian-Far, R. Montasari, A. Daneshkhah

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

14 Citations (Scopus)

100 Downloads (Pure)

Abstract

The high velocity, variety and volume of data generation by today's systems have necessitated Big Data (BD) analytic techniques. This has penetrated a wide range of industries; BD as a notion has various types and characteristics, and therefore a variety of analytic techniques would be required. The traditional analysis methods are typically unable to analyse spatial-temporal BD. Interpolation is required to approximate the values between the already existing data points, yet since there exist both location and time dimensions, only a multivariate interpolation would be appropriate. Nevertheless, existing software are unable to perform such complex interpolations. To overcome this challenge, this paper presents a layer by layer interpolation approach for spatial-temporal BD. Developing this layered structure provides the opportunity for working with much smaller linear system of equations. Consequently, this structure increases the accuracy and stability of numerical structure of the considered BD interpolation. To construct this layer by layer interpolation, we have used the good properties of Radial Basis Functions (RBFs). The proposed new approach is applied to numerical examples in spatial-temporal big data and the obtained results confirm the high accuracy and low computational cost. Finally, our approach is applied to explore one of the air pollution indices, i.e. daily PM_2.5 concentration, based on different stations in the contiguous United States, and it is evaluated by leave-one-out cross validation.

Original language	English
Pages (from-to)	492-502
Number of pages	11
Journal	Applied Numerical Mathematics
Volume	153
Early online date	13 Mar 2020
DOIs	https://doi.org/10.1016/j.apnum.2020.03.009
Publication status	Published - 1 Jul 2020

Bibliographical note

NOTICE: this is the author’s version of a work that was accepted for publication in Applied Numerical Mathematics. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Applied Numerical Mathematics, 153, (2020) DOI: 10.1016/j.apnum.2020.03.009

© 2020, Elsevier. Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/

Funder

UK Engineering & Physical Sciences Research Council (Strategic Package: Centre for Predictive Modelling in Science and Engineering - Grant No. EP/L027682/1)

Keywords

Big data
Layer by layer interpolation
Radial basis functions
Spatial-temporal interpolation

ASJC Scopus subject areas

Numerical Analysis
Computational Mathematics
Applied Mathematics

Access to Document

10.1016/j.apnum.2020.03.009

Post-PrintAccepted author manuscript, 706 KBLicence: CC BY-NC-ND

Cite this

@article{329ffac595764804b842e79b0abc913b,

title = "A low cost and highly accurate technique for big data spatial-temporal interpolation",

abstract = "The high velocity, variety and volume of data generation by today's systems have necessitated Big Data (BD) analytic techniques. This has penetrated a wide range of industries; BD as a notion has various types and characteristics, and therefore a variety of analytic techniques would be required. The traditional analysis methods are typically unable to analyse spatial-temporal BD. Interpolation is required to approximate the values between the already existing data points, yet since there exist both location and time dimensions, only a multivariate interpolation would be appropriate. Nevertheless, existing software are unable to perform such complex interpolations. To overcome this challenge, this paper presents a layer by layer interpolation approach for spatial-temporal BD. Developing this layered structure provides the opportunity for working with much smaller linear system of equations. Consequently, this structure increases the accuracy and stability of numerical structure of the considered BD interpolation. To construct this layer by layer interpolation, we have used the good properties of Radial Basis Functions (RBFs). The proposed new approach is applied to numerical examples in spatial-temporal big data and the obtained results confirm the high accuracy and low computational cost. Finally, our approach is applied to explore one of the air pollution indices, i.e. daily PM2.5 concentration, based on different stations in the contiguous United States, and it is evaluated by leave-one-out cross validation.",

keywords = "Big data, Layer by layer interpolation, Radial basis functions, Spatial-temporal interpolation",

author = "M. Esmaeilbeigi and O. Chatrabgoun and A. Hosseinian-Far and R. Montasari and A. Daneshkhah",

note = "NOTICE: this is the author{\textquoteright}s version of a work that was accepted for publication in Applied Numerical Mathematics. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Applied Numerical Mathematics, 153, (2020) DOI: 10.1016/j.apnum.2020.03.009 {\textcopyright} 2020, Elsevier. Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/ ",

year = "2020",

month = jul,

day = "1",

doi = "10.1016/j.apnum.2020.03.009",

language = "English",

volume = "153",

pages = "492--502",

journal = "Applied Numerical Mathematics",

issn = "0168-9274",

publisher = "Elsevier",

}

TY - JOUR

T1 - A low cost and highly accurate technique for big data spatial-temporal interpolation

AU - Esmaeilbeigi, M.

AU - Chatrabgoun, O.

AU - Hosseinian-Far, A.

AU - Montasari, R.

AU - Daneshkhah, A.

N1 - NOTICE: this is the author’s version of a work that was accepted for publication in Applied Numerical Mathematics. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reflected in this document. Changes may have been made to this work since it was submitted for publication. A definitive version was subsequently published in Applied Numerical Mathematics, 153, (2020) DOI: 10.1016/j.apnum.2020.03.009 © 2020, Elsevier. Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International http://creativecommons.org/licenses/by-nc-nd/4.0/

PY - 2020/7/1

Y1 - 2020/7/1

N2 - The high velocity, variety and volume of data generation by today's systems have necessitated Big Data (BD) analytic techniques. This has penetrated a wide range of industries; BD as a notion has various types and characteristics, and therefore a variety of analytic techniques would be required. The traditional analysis methods are typically unable to analyse spatial-temporal BD. Interpolation is required to approximate the values between the already existing data points, yet since there exist both location and time dimensions, only a multivariate interpolation would be appropriate. Nevertheless, existing software are unable to perform such complex interpolations. To overcome this challenge, this paper presents a layer by layer interpolation approach for spatial-temporal BD. Developing this layered structure provides the opportunity for working with much smaller linear system of equations. Consequently, this structure increases the accuracy and stability of numerical structure of the considered BD interpolation. To construct this layer by layer interpolation, we have used the good properties of Radial Basis Functions (RBFs). The proposed new approach is applied to numerical examples in spatial-temporal big data and the obtained results confirm the high accuracy and low computational cost. Finally, our approach is applied to explore one of the air pollution indices, i.e. daily PM2.5 concentration, based on different stations in the contiguous United States, and it is evaluated by leave-one-out cross validation.

AB - The high velocity, variety and volume of data generation by today's systems have necessitated Big Data (BD) analytic techniques. This has penetrated a wide range of industries; BD as a notion has various types and characteristics, and therefore a variety of analytic techniques would be required. The traditional analysis methods are typically unable to analyse spatial-temporal BD. Interpolation is required to approximate the values between the already existing data points, yet since there exist both location and time dimensions, only a multivariate interpolation would be appropriate. Nevertheless, existing software are unable to perform such complex interpolations. To overcome this challenge, this paper presents a layer by layer interpolation approach for spatial-temporal BD. Developing this layered structure provides the opportunity for working with much smaller linear system of equations. Consequently, this structure increases the accuracy and stability of numerical structure of the considered BD interpolation. To construct this layer by layer interpolation, we have used the good properties of Radial Basis Functions (RBFs). The proposed new approach is applied to numerical examples in spatial-temporal big data and the obtained results confirm the high accuracy and low computational cost. Finally, our approach is applied to explore one of the air pollution indices, i.e. daily PM2.5 concentration, based on different stations in the contiguous United States, and it is evaluated by leave-one-out cross validation.

KW - Big data

KW - Layer by layer interpolation

KW - Radial basis functions

KW - Spatial-temporal interpolation

UR - http://www.scopus.com/inward/record.url?scp=85081885549&partnerID=8YFLogxK

U2 - 10.1016/j.apnum.2020.03.009

DO - 10.1016/j.apnum.2020.03.009

M3 - Article

AN - SCOPUS:85081885549

SN - 0168-9274

VL - 153

SP - 492

EP - 502

JO - Applied Numerical Mathematics

JF - Applied Numerical Mathematics

ER -

A low cost and highly accurate technique for big data spatial-temporal interpolation

Abstract

Bibliographical note

Funder

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this