Abstract
Many missing data imputation methods are based on only complete instances (instances without missing values in a dataset) when estimating plausible values for the missing values in the dataset. Actually, the information within incomplete instances (instances with missing values) can also play an important role in missing value imputation. For example, the information has been applied to identifying the neighbors of an instance with missing values in NN (nearest neighbor) imputation, and the class of the instance in clustering-based imputation, where NN and clustering-based imputations are well-known efficient algorithms. Therefore, in this paper we advocate to well utilize the information within incomplete instances when estimating missing values. As an attempt, a simple and efficient nonparametric iterative imputation algorithm, called NIIA method, is designed for imputing iteratively missing target values. The NIIA method imputes each missing value several times until the algorithm converges. In the first iteration, all complete instances are used to estimate missing values. The information within incomplete instances is utilized since the second iteration. We conduct intensive experiments for evaluating the proposed approach. Our experimental results show: (1) The utilization of information within incomplete instances is of benefit to capture the distribution of a dataset much better and easier than parametric imputation. (2) NIIA method outperforms the existing methods at the accuracy, and this advantage is clearly highlighted when datasets are with high missing ratio.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Blake, C., Merz, C.: UCI Repository of machine learning databases (1998)
Caruana, R.: A Non-parametric EM-style algorithm for Imputing Missing Value. Artificial Intelligence and Statistics (January 2001)
Little, R., Rubin, D.: Statistical Analysis with Missing Data, 2nd edn. John Wiley and Sons, New York (2002)
Pearson, P.K.: Mining imperfect data: dealing with contamination and incomplete records. SIAM, Philadelphia (2005)
Wang, Q., Rao, J.N.K.: Empirical likelihood-based inference under imputation for missing response data. Ann. Statist. 30, 896–924 (2002)
Zhang, C.Q., et al.: Efficient Imputation Method for Missing Values. In: Zhou, Z.-H., Li, H., Yang, Q. (eds.) PAKDD 2007. LNCS (LNAI), vol. 4426, pp. 1080–1087. Springer, Heidelberg (2007)
Qin, Y.S., et al.: Optimized parameters for missing data imputation. In: Yang, Q., Webb, G. (eds.) PRICAI 2006. LNCS (LNAI), vol. 4099, pp. 1010–1016. Springer, Heidelberg (2006)
Qin, Y.S., et al.: POP Algorithm: Kernel-Based Imputation to Treat Missing Values in Knowledge Discovery from Databases. Expert Systems with Applications (accepted, 2008), http://dx.doi.org/10.1016/j.eswa.2008.01.059
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2008 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Zhang, S., Jin, Z., Zhu, X. (2008). NIIA: Nonparametric Iterative Imputation Algorithm. In: Ho, TB., Zhou, ZH. (eds) PRICAI 2008: Trends in Artificial Intelligence. PRICAI 2008. Lecture Notes in Computer Science(), vol 5351. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-89197-0_50
Download citation
DOI: https://doi.org/10.1007/978-3-540-89197-0_50
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-89196-3
Online ISBN: 978-3-540-89197-0
eBook Packages: Computer ScienceComputer Science (R0)