A Google Trends-Based Spatial Weighting Approach for Space-Time Modeling of Gold Prices during the COVID-19 Pandemic

Authors

DOI:

https://doi.org/10.24252/msa.v14i2.62471

Keywords:

forecasting, GSTARX, GSTARIX, imputation technique, spatial weighting matrix

Abstract

This study aims to forecast gold prices in Austria, India, South Korea, and Turkiye during the COVID-19 pandemic using an optimal space-time modeling approach. Previous studies have primarily employed conventional spatial weighting matrices based on geographical proximity or economic relationships, with limited attention given to utilizing online search behavior as a source of spatial dependence, particularly when handling incomplete data. To address this research gap, this study proposes a novel framework that integrates Google Trends data as both a data-driven spatial weighting matrix and an exogenous variable within generalized space-time autoregressive (GSTAR)-based models. Missing observations were first imputed using the next observation carried backward (NOCB) method before model estimation. Subsequently, GSTARX and GSTARIX models were developed using Google Trends data from five search categories. Model performance was evaluated using the mean absolute percentage error (MAPE). The results demonstrate that the GSTARX model with a spatial weighting matrix derived from Google Images achieved the highest forecasting accuracy, yielding the lowest MAPE of 4.3%. The forecasts indicate relatively stable gold prices in Austria, declining trends in India and South Korea with a more pronounced decline in South Korea and a substantial increase in Turkiye. These findings demonstrate that Google Trends data can effectively capture dynamic spatial relationships beyond conventional weighting schemes, offering a practical and data-driven alternative for improving space-time forecasting accuracy in the presence of missing values.

References

M. A. Novianto, Suhartono, D. D. Prastyo, A. Suharsono, and Setiawan, “GSTARIX model for forecasting spatio-temporal data with trend, seasonal and intervention,” J. Phys. Conf. Ser., vol. 1097, p. 012076, Sep. 2018, doi: 10.1088/1742-6596/1097/1/012076.

P. Monika, B. N. Ruchjana, and A. S. Abdullah, “GSTARI-X-ARCH model with data mining approach for forecasting climate in West Java,” Computation, vol. 10, no. 12, p. 204, Nov. 2022, doi: 10.3390/computation10120204.

E. Setyowati, Suhartono, and D. D. Prastyo, “A hybrid generalized space-time autoregressive-Elman recurrent neural network model for forecasting space-time data with exogenous variables,” J. Phys. Conf. Ser., vol. 1752, no. 1, p. 012012, Feb. 2021, doi: 10.1088/1742-6596/1752/1/012012.

N. M. Huda and N. Imro’ah, “Determination of the best weight matrix for the generalized space time autoregressive (GSTAR) model in the COVID-19 case on Java Island, Indonesia,” Spat. Stat., vol. 54, p. 100734, Apr. 2023, doi: 10.1016/j.spasta.2023.100734.

U. Mukhaiyar, A. W. Mahdiyasa, K. N. Sari, and N. T. Noviana, “The generalized STAR modeling with minimum spanning tree approach of spatial weight matrix,” Front. Appl. Math. Stat., vol. 10, Aug. 2024, doi: 10.3389/fams.2024.1417037.

A. Ashari, A. Efendi, and H. Pramoedyo, “GSTARX-SUR modeling using inverse distance weighted matrix and queen contiguity weighted matrix for forecasting cocoa black pod attack in Trenggalek Regency,” in Proceedings of the Proceedings of the 13th International Interdisciplinary Studies Seminar, IISS 2019, 30-31 October 2019, Malang, Indonesia, EAI, 2020. doi: 10.4108/eai.23-10-2019.2293086.

K. Khadijah, H. Hikmah, F. Fardinah, A. Apriyanto, N. Nurhidayah, and S. D. Lestari, “Model generalized space time autoregressive with variable exogenous (GSTARX) dalam meramalkan data inflasi di Sulawesi Selatan,” Journal of Mathematics: Theory and Applications, vol. 6, no. 1, pp. 8–18, Apr. 2024, doi: 10.31605/jomta.v6i1.2999.

V. Z. Eichenauer, R. Indergand, I. Z. Martínez, and C. Sax, “Obtaining consistent time series from Google Trends,” Econ. Inq., vol. 60, no. 2, pp. 694–705, Apr. 2022, doi: 10.1111/ecin.13049.

J. Hölzl, F. Keusch, and C. Sajons, “The (mis)use of Google Trends data in the social sciences - A systematic review, critique, and recommendations,” Soc. Sci. Res., vol. 126, p. 103099, Feb. 2025, doi: 10.1016/j.ssresearch.2024.103099.

T. Hu et al., “Human mobility data in the COVID-19 pandemic: Characteristics, applications, and challenges,” Int. J. Digit. Earth, vol. 14, no. 9, pp. 1126–1147, Sep. 2021, doi: 10.1080/17538947.2021.1952324.

C. Ilin, S. Annan-Phan, X. H. Tai, S. Mehra, S. Hsiang, and J. E. Blumenstock, “Public mobility data enables COVID-19 forecasting and management at local and global scales,” Sci. Rep., vol. 11, no. 1, p. 13531, Jun. 2021, doi: 10.1038/s41598-021-92892-8.

H. Kwon and C. Koylu, “Revealing associations between spatial time series trends of COVID-19 incidence and human mobility: an analysis of bidirectionality and spatiotemporal heterogeneity,” Int. J. Health Geogr., vol. 22, no. 1, p. 33, Nov. 2023, doi: 10.1186/s12942-023-00357-0.

A. Potgieter et al., “Modelling representative population mobility for COVID-19 spatial transmission in South Africa,” Front. Big Data, vol. 4, Oct. 2021, doi: 10.3389/fdata.2021.718351.

W. Ming et al., “Understanding health communication through Google Trends and news coverage for COVID-19: Multinational study in eight countries,” JMIR Public Health Surveill., vol. 7, no. 12, p. e26644, Dec. 2021, doi: 10.2196/26644.

[15] D. Wang, J. C. Lang, and Y.-H. Chen, “Assessment of using Google Trends for real-time monitoring of infectious disease outbreaks: A measles case study,” Sci. Rep., vol. 14, no. 1, p. 9470, Apr. 2024, doi: 10.1038/s41598-024-601208.

C. Wychgram, J. N. Aucott, A. W. Rebman, and F. C. Curriero, “Identifying the geographic leading edge of Lyme disease in the United States with internet searches: A spatiotemporal analysis of Google Health Trends data,” PLoS One, vol. 19, no. 11, p. e0312277, Nov. 2024, doi: 10.1371/journal.pone.0312277.

Downloads

Published

2026-10-21

How to Cite

[1]
F. Mubarak, V. Y. Sundara, Nurniswah, Germansah, and P. Jiblathar, “A Google Trends-Based Spatial Weighting Approach for Space-Time Modeling of Gold Prices during the COVID-19 Pandemic”, MSA, vol. 14, no. 2, pp. 214–223, Oct. 2026.

Issue

Vol. 14 No. 2 (2026): Volume 14 No 2, 2026

Section

Artikel

License

Copyright (c) 2026 Jurnal MSA ( Matematika dan Statistika serta Aplikasinya)

Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Copyright Notice

Jurnal MSA (Matematika dan Statistika serta Aplikasinya)

Authors who publish in Jurnal MSA agree to the following copyright and licensing terms.

1. Copyright Ownership

Authors retain copyright in their scholarly articles unless copyright is lawfully held by another authorized rights holder.

Publication in Jurnal MSA does not require an exclusive transfer or assignment of copyright ownership to the journal or its publisher.

2. Non-Exclusive Publishing Rights

Authors or authorized copyright holders grant Jurnal MSA non-exclusive rights of first journal publication, dissemination, indexing, and digital preservation.

These rights allow the journal to publish, reproduce, distribute, archive, and preserve the accepted scholarly article without transferring copyright ownership.

3. Publication License

Jurnal MSA adopts and applies the Creative Commons Attribution 4.0 International License (CC BY 4.0) as the standard license for scholarly articles published under its current publication terms.

Authors or authorized rights holders authorize publication of their articles under CC BY 4.0.

This license permits sharing, reproduction, redistribution, adaptation, and reuse, including commercial reuse, subject to the license terms and attribution requirements.

Official License:

https://creativecommons.org/licenses/by/4.0/

4. Non-Commercial Publishing Model

Jurnal MSA operates as a non-commercial scholarly publisher dedicated to education, scientific research, and scholarly communication.

The journal's non-commercial status does not restrict commercial reuse of articles licensed under CC BY 4.0.

5. Authors' Rights and Self-Archiving

Authors retain their applicable rights to reuse, share, and disseminate their scholarly works.

Authors may immediately deposit and distribute the published version of their CC BY 4.0-licensed articles through institutional repositories, disciplinary repositories, and academic websites without an embargo imposed by the journal.

Such distribution remains subject to the applicable license and third-party rights.

6. Author Consent and Third-Party Rights

By agreeing to publication, the corresponding author confirms that:

  1. All listed authors have approved the manuscript and the publication arrangements.

  2. The authors or authorized rights holders possess the legal authority necessary to authorize publication and licensing.

  3. Jurnal MSA is granted the necessary non-exclusive publishing rights.

  4. Publication of the article under CC BY 4.0 is authorized.

  5. Required permissions for third-party materials have been obtained, and any licensing exclusions have been clearly identified.

7. Previously Published Articles

Articles published before the implementation of the current licensing policy remain governed by their validly applicable copyright and licensing arrangements.

The adoption of CC BY 4.0 does not automatically change licenses previously granted under different terms.

Copyright and License Statement

Copyright © [Year], The Author(s).

The author(s) retain copyright and grant Jurnal MSA non-exclusive rights of first publication, dissemination, indexing, and preservation.

This article is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).

https://creativecommons.org/licenses/by/4.0/