A Comparison of Five Methods for Missing Value Imputation in Data Sets

Öz

The missing values in the data sets do not allow for accurate analysis. Therefore, the correct imputation of missing values has become the focus of attention of researchers in recent years. This paper focuses on a comparison of most reliable and up to date estimation methods to imputing the missing values. Imputation of missing values has a very high priority because of its impact on next pre-processing, data analysis, classification, clustering, etc. Root mean square error (RMSE) value, classification accuracy and execution time are used to evaluate the performances of most popular five methods (mean, k-nearest neighbors, singular value decomposition, bayesian principal component analysis and missForest). When RMSE and classification accuracy values of methods were compared, it has observed that missForest method outperformed other methods in all datasets.

Anahtar Kelimeler

• Missing value imputation,k-nearest neighbor,singular value decomposition,bayesian principal component analysis,missForest

A Comparison of Five Methods for Missing Value Imputation in Data Sets

Öz

The missing values in the data sets do not allow for accurate analysis. Therefore, the correct imputation of missing values has become the focus of attention of researchers in recent years. This paper focuses on a comparison of most reliable and up to date estimation methods to imputing the missing values. Imputation of missing values has a very high priority because of its impact on next pre-processing, data analysis, classification, clustering, etc. Root mean square error (RMSE) value, classification accuracy and execution time are used to evaluate the performances of most popular five methods (mean, k-nearest neighbors, singular value decomposition, bayesian principal component analysis and missForest). When RMSE and classification accuracy values of methods were compared, it has observed that missForest method outperformed other methods in all datasets.

Anahtar Kelimeler

• Missing value imputation,k-nearest neighbor,singular value decomposition,bayesian principal component analysis,missForest

Kaynakça

1. [1] T.D. Pigott, “A review of methods for missing data”, Educational Resarch and Evaluation, Cilt. 7, s. 353-383. DOI: 10.1076/edre.7.4.353.8937, 2001.
2. [2] P.D. Allison, “Missing data techniques for structural equation modeling”, Journal of Abnormal Psychology, Cilt. 4, s. 545-557. DOI: 10.1037/0021-843X.112.4.545, 2003.
3. [3] J.W. Osborne, “Best practices in data cleaning”, California: Sage Publication, Inc., s. 596, 2013.
4. [4] A.G. Di Nuovo, “Missing data analysis with fuzzy C-Means: A study of its application in a psychological scenario”, Expert Syst Appl, Cilt. 38, s. 6793-6797, DOI: 10.1016/j.eswa.2010.12.067, 2011.
5. [5] C. Bergmeir, J.M. Benitez, “On the use of cross-validation for time series predictor evaluation”, Inform Sciences, Cilt. 191, s. 192-213, DOI: 10.1016/j.ins.2011.12.028, 2012.
6. [6] J. Van Hulse, and T.M. Khoshgoftaar, “Incomplete-case nearest neighbor imputation in software measurement data”, IRI 2007: Proceedings of the 2007 IEEE International Conference on Information Reuse and Integration, s. 630-637 DOI: 10.1109/IRI.2007.4296691, 2007.
7. [7] S. Genc, F.E.Boran, D. Akay, and Z.S. Xu, “Interval multiplicative transitivity for consistency, missing values and priority weights of interval fuzzy preference relations”, Inform Sciences, Cilt. 180, s. 4877-4891, DOI: 10.1016/j.ins.2010.08.019, 2010.
8. [8] R.J.A. Little, and D.B. Rubin, “Statistical Analysis with Missing Data”, 333. John Wiley & Sons, 2014.

9. [9] O. Troyanskaya, M. Cantor, G. Sherlock, P. Brown, T. Hastie, vd. “Missing value estimation methods for dna microarrays”, Bioinformatics, Cilt. 17, s. 520-525, 2001.
10. [10] S. Oba, M.A. Sato, I. Takemasa, M. Monden, K.I. Matsubara, and S. Ishii, “A bayesian missing value estimation method for gene expression profile data”, Bioinformatics, Cilt. 19, s. 2088-2096, DOI: 10.1093/bioinformatics/btg287, 2003.
11. [11] D.J. Stekhoven, and P. Buhlmann, “Miss Forest - nonparametric missing value imputation for mixed-type data”, Bioinformatics. Cilt. 28, s. 112-118, DOI: 10.1093/bioinformatics/btr597, 2012.
12. [12] T. Marwala, “Computational intelligence for missing data imputation, estimation and management:knowledge optimization techniques”, Information Science Reference, Hershey PA, 2009.
13. [13] P. Cihan, “Veri madenciliği yöntemleriyle hayvan hastalıklarında teşhis, prognoz ve risk faktörlerinin belirlenmesi”. Yıldız Teknik Üniversitesi, Fen Bilimleri Enstitüsü, Doktora Tezi, 101s, İstanbul, 2018.
14. [14] Lichman, M. UCI Machine Learning Repository, http:// archive.ics.uci.edu/ml (Erişim Tarihi: 15.03.2018).
15. [15] L. Breiman, Random forests, in: Machine Learning, Kluwer Academic Publishers, 2001.
16. [16] R.S. Marko, “Improving random forests”, In: European conference on machine learning, Springer, Berlin, Heidelberg, 2004.
17. [17] H. Kim, H. Kim, H. Moon, and H. Ahn, “A weight-adjusted voting algorithm for ensembles of classifiers”, J. Korean Stat. Soc., Cilt. 40, s. 437-449, DOI: 10.1016/j.jkss.2011.03.002, 2011.
18. [18] H.B. Li, W. Wang, H.W. Ding, and J. Dong, “Trees weighting random forest method for classifying high-dimensional noisy data”, e-Business Engineering (ICEBE) in: 2010 IEEE 7th International Conference on IEEE, s. 160–163, 2010.