Missing values imputation in time series using decision trees
DOI:
https://doi.org/10.5902/2179460X84257Keywords:
ARMA, Random walk, Decision trees, Missing data, ImputationAbstract
Filling in missing values in time series is a problem that has received little attention. The studies found in the literature generally focus on linear models from the ARIMA family and do not discuss the validity of proposed methodologies for cases with a large volume of missing data, in which parametric methods become challenging due to the additional problem of identifying the order of the model. To address these issues, this study proposes a methodology for time series reconstruction using decision trees, a machine learning method that does not assume a parametric model for the data. In this approach, the known values of the time series act as the response variable, while corresponding lags are used as predictors. The tree selected by the training algorithm is then used to predict the missing values in the response. Monte Carlo simulations are used to investigate the proposed methodology, considering processes from the ARMA family and the random walk while varying the size of the time series, model parameters, proportion of missing values, and the predictors. To evaluate the quality of the reconstructions, the predictions of the decision trees are compared with those of some traditional imputation methods. The results demonstrate the potential of the proposed method and are consistent with the theoretical framework of this study. To promote the proposed methodology, a shiny application has been developed and made publicly available.
Downloads
References
Batista, G., Monard, M. C. (2003). An analysis of four missing data treatment methods for supervised learning. Applied Artificial Intelligence, 17, 519–533.
Breiman, L., Friedman, J., Stone, C., Olshen, R. (1984). Classification and Regression Trees. Taylor & Francis.
Brockwell, P. J., Davis, R. A. (1991). Time Series: Theory and Methods, 2o ed. Springer Science & Business Media.
Chang, W., Cheng, J., Allaire, J., Sievert, C., Schloerke, B., Xie, Y., Allen, J., McPherson, J., Dipert, A., Borges, B. (2021). shiny: Web Application Framework for R. URL https://CRAN.R-project.org/package=shiny , r package version 1.7.1.
Dempster, A. P., Laird, N. M., Rubin, D. B. (1977). Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society Series B (Methodological), 39(1), 1–38.
Dergachev, V. A., Gorban, A. N., Rossiev, A. A., Karimova, L. M., Kuandykov, E. B., Makarenko, N. G., Steier, P. (2001). The filling of gaps in geophysical time series by artificial neural networks. Radiocarbon, 43(2A), 365–371.
Greiner, R., Grove, A., Kogan, A. (1997). Knowing what doesn’t matter: exploiting the omission of irrelevant data. Artificial Intelligence, 97(1-2), 345–380.
Hastie, T., Tibshirani, R., Friedman, J. (2009). The elements of statistical learning: data mining, inference, and prediction. Springer Science & Business Media.
James, G., Witten, D., Hastie, T., Tibshirani, R. (2013). An Introduction to Statistical Learning: with Applications in R. Springer Texts in Statistics, Springer New York.
Josse, J., Prost, N., Scornet, E., Varoquaux, G. (2019). On the consistency of supervised learning with missing values. arXiv:190206931.
Kass, G. V. (1980). An exploratory technique for investigating large quantities of categorical data. Applied Statistics, 20(2), 119–127.
Little, R. J. A. (1995). Modeling the drop-out mechanism in repeated-measures studies. Journal of the American Statistical Association, 90(431), 1112–1121.
Ljung, G. M. (1989). A note on the estimation of missing values in time series. Communications in Statistics - Simulation and Computation, 18(2), 459–465.
Luceño, A. (1997). Estimation of missing values in possibly partially nonstationary vector time series. Biometrika, 84(2), 495–499.
Molenberghs, G., Fitzmaurice, G. M., Kenward, M. G., Tsiatis, A. A., Verbeke, G. (2020). Handbook of Missing Data Methodology. Chapman & Hall/CRC Handbooks of Modern Statistical Methods, Taylor & Francis Group.
Morettin, P. A., Toloi, C. M. d. C. (2004). Análise de séries temporais. Edgard Blucher.
Moritz, S., Bartz-Beielstein, T. (2017). imputeTS: Time Series Missing Value Imputation in R. The R Journal, 9(1), 207–218.
Murphy, K. (2012). Machine Learning: A Probabilistic Perspective. Adaptive Computation and Machine Learning series, MIT Press.
Prass, T. S., Pumi, G. (2021). On the behavior of the DFA and DCCA in trend-stationary processes. Journal of Multivariate Analysis, 182, 104,703.
Pratama, I., Permanasari, A., Ardiyanto, I., Indrayani, R. (2016). A review of missing values handling methods on time-series data. Em: 2016 International Conference on Information Technology Systems and Innovation (ICITSI), pp. 1–6.
Quinlan, J. R. (1986). Induction of decision trees. Machine Learning, 1(1), 81–106.
R Core Team (2022). R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, URL https://www.R-project.org/.
RStudio Team (2022). RStudio: Integrated Development Environment for R. RStudio, PBC, Boston, MA, URL http://www.rstudio.com/.
Rubin, D. B. (1976). Inference and missing data. Biometrika, 63, 581–592.
Shumway, R. H., Stoffer, D. S. (2005). Time Series Analysis and Its Applications (Springer Texts in Statistics). Springer-Verlag, Berlin, Heidelberg.
Therneau, T., Atkinson, B. (2019). rpart: Recursive Partitioning and Regression Trees. URL https://CRAN.R-project. org/package=rpart, r package version 4.1-15.
Van der Vaart, A. W. (2010). Time series. Lecture notes for courses “Tijdreeksen”, “Time Series” and “Financial Time Series” held at Vrije Universiteit Amsterdam, 1995-2010. URL https://staff.fnwi.uva.nl/p.j.c.spreij/onderwijs/master/aadtimeseries2010.pdf.
Yodah, Kihoro, J., Athiany, H., W, W., Kibunja (2013). Imputation of incomplete non-stationary seasonal time series data. Mathematical Theory and Modeling, 3, 142–154.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2024 Ciência e Natura
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
To access the DECLARATION AND TRANSFER OF COPYRIGHT AUTHOR’S DECLARATION AND COPYRIGHT LICENSE click here.
Ethical Guidelines for Journal Publication
The Ciência e Natura journal is committed to ensuring ethics in publication and quality of articles.
Conformance to standards of ethical behavior is therefore expected of all parties involved: Authors, Editors, Reviewers, and the Publisher.
In particular,
Authors: Authors should present an objective discussion of the significance of research work as well as sufficient detail and references to permit others to replicate the experiments. Fraudulent or knowingly inaccurate statements constitute unethical behavior and are unacceptable. Review Articles should also be objective, comprehensive, and accurate accounts of the state of the art. The Authors should ensure that their work is entirely original works, and if the work and/or words of others have been used, this has been appropriately acknowledged. Plagiarism in all its forms constitutes unethical publishing behavior and is unacceptable. Submitting the same manuscript to more than one journal concurrently constitutes unethical publishing behavior and is unacceptable. Authors should not submit articles describing essentially the same research to more than one journal. The corresponding Author should ensure that there is a full consensus of all Co-authors in approving the final version of the paper and its submission for publication.
Editors: Editors should evaluate manuscripts exclusively on the basis of their academic merit. An Editor must not use unpublished information in the editor's own research without the express written consent of the Author. Editors should take reasonable responsive measures when ethical complaints have been presented concerning a submitted manuscript or published paper.
Reviewers: Any manuscripts received for review must be treated as confidential documents. Privileged information or ideas obtained through peer review must be kept confidential and not used for personal advantage. Reviewers should be conducted objectively, and observations should be formulated clearly with supporting arguments, so that Authors can use them for improving the paper. Any selected Reviewer who feels unqualified to review the research reported in a manuscript or knows that its prompt review will be impossible should notify the Editor and excuse himself from the review process. Reviewers should not consider manuscripts in which they have conflicts of interest resulting from competitive, collaborative, or other relationships or connections with any of the authors, companies, or institutions connected to the papers.
