Scalable model-based cascaded imputation of missing data

Jacob Montiel; Jesse Read; Albert Bifet; Talel Abdessalem

doi:10.1007/978-3-319-93040-4_6

Scalable model-based cascaded imputation of missing data

Jacob Montiel, Jesse Read, Albert Bifet, Talel Abdessalem

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Missing data is a common trait of real-world data that can negatively impact interpretability. In this paper, we present Cascade Imputation (CIM), an effective and scalable technique for automatic imputation of missing data. CIM is not restrictive on the characteristics of the data set, providing support for: Missing At Random and Missing Completely At Random data, numerical and nominal attributes, and large data sets including highly dimensional data sets. We compare CIM against well-established imputation techniques over a variety of data sets under multiple test configurations to measure the impact of imputation on the classification problem. Test results show that CIM outperforms other imputation methods over multiple test conditions. Additionally, we identify optimal performance and failure conditions for popular imputation techniques.

Original language	English
Title of host publication	Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings
Editors	Geoffrey I. Webb, Dinh Phung, Mohadeseh Ganji, Lida Rashidi, Vincent S. Tseng, Bao Ho
Publisher	Springer Verlag
Pages	64-76
Number of pages	13
ISBN (Print)	9783319930398
DOIs	https://doi.org/10.1007/978-3-319-93040-4_6
Publication status	Published - 1 Jan 2018
Externally published	Yes
Event	22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018 - Melbourne, Australia Duration: 3 Jun 2018 → 6 Jun 2018

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10939 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018
Country/Territory	Australia
City	Melbourne
Period	3/06/18 → 6/06/18

Keywords

Classification
Imputation
Missing data

Access to Document

10.1007/978-3-319-93040-4_6

Cite this

Montiel, J., Read, J., Bifet, A., & Abdessalem, T. (2018). Scalable model-based cascaded imputation of missing data. In G. I. Webb, D. Phung, M. Ganji, L. Rashidi, V. S. Tseng, & B. Ho (Eds.), Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings (pp. 64-76). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10939 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-93040-4_6

Montiel, Jacob ; Read, Jesse ; Bifet, Albert et al. / Scalable model-based cascaded imputation of missing data. Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. editor / Geoffrey I. Webb ; Dinh Phung ; Mohadeseh Ganji ; Lida Rashidi ; Vincent S. Tseng ; Bao Ho. Springer Verlag, 2018. pp. 64-76 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{291472ce7d454e60bb474b825a9c99aa,

title = "Scalable model-based cascaded imputation of missing data",

abstract = "Missing data is a common trait of real-world data that can negatively impact interpretability. In this paper, we present Cascade Imputation (CIM), an effective and scalable technique for automatic imputation of missing data. CIM is not restrictive on the characteristics of the data set, providing support for: Missing At Random and Missing Completely At Random data, numerical and nominal attributes, and large data sets including highly dimensional data sets. We compare CIM against well-established imputation techniques over a variety of data sets under multiple test configurations to measure the impact of imputation on the classification problem. Test results show that CIM outperforms other imputation methods over multiple test conditions. Additionally, we identify optimal performance and failure conditions for popular imputation techniques.",

keywords = "Classification, Imputation, Missing data",

author = "Jacob Montiel and Jesse Read and Albert Bifet and Talel Abdessalem",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG, part of Springer Nature 2018.; 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018 ; Conference date: 03-06-2018 Through 06-06-2018",

year = "2018",

month = jan,

day = "1",

doi = "10.1007/978-3-319-93040-4\_6",

language = "English",

isbn = "9783319930398",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "64--76",

editor = "Webb, \{Geoffrey I.\} and Dinh Phung and Mohadeseh Ganji and Lida Rashidi and Tseng, \{Vincent S.\} and Bao Ho",

booktitle = "Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings",

}

Montiel, J, Read, J, Bifet, A & Abdessalem, T 2018, Scalable model-based cascaded imputation of missing data. in GI Webb, D Phung, M Ganji, L Rashidi, VS Tseng & B Ho (eds), Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10939 LNAI, Springer Verlag, pp. 64-76, 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018, Melbourne, Australia, 3/06/18. https://doi.org/10.1007/978-3-319-93040-4_6

Scalable model-based cascaded imputation of missing data. / Montiel, Jacob; Read, Jesse; Bifet, Albert et al.
Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. ed. / Geoffrey I. Webb; Dinh Phung; Mohadeseh Ganji; Lida Rashidi; Vincent S. Tseng; Bao Ho. Springer Verlag, 2018. p. 64-76 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10939 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Scalable model-based cascaded imputation of missing data

AU - Montiel, Jacob

AU - Read, Jesse

AU - Bifet, Albert

AU - Abdessalem, Talel

N1 - Publisher Copyright: © Springer International Publishing AG, part of Springer Nature 2018.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Missing data is a common trait of real-world data that can negatively impact interpretability. In this paper, we present Cascade Imputation (CIM), an effective and scalable technique for automatic imputation of missing data. CIM is not restrictive on the characteristics of the data set, providing support for: Missing At Random and Missing Completely At Random data, numerical and nominal attributes, and large data sets including highly dimensional data sets. We compare CIM against well-established imputation techniques over a variety of data sets under multiple test configurations to measure the impact of imputation on the classification problem. Test results show that CIM outperforms other imputation methods over multiple test conditions. Additionally, we identify optimal performance and failure conditions for popular imputation techniques.

AB - Missing data is a common trait of real-world data that can negatively impact interpretability. In this paper, we present Cascade Imputation (CIM), an effective and scalable technique for automatic imputation of missing data. CIM is not restrictive on the characteristics of the data set, providing support for: Missing At Random and Missing Completely At Random data, numerical and nominal attributes, and large data sets including highly dimensional data sets. We compare CIM against well-established imputation techniques over a variety of data sets under multiple test configurations to measure the impact of imputation on the classification problem. Test results show that CIM outperforms other imputation methods over multiple test conditions. Additionally, we identify optimal performance and failure conditions for popular imputation techniques.

KW - Classification

KW - Imputation

KW - Missing data

U2 - 10.1007/978-3-319-93040-4_6

DO - 10.1007/978-3-319-93040-4_6

M3 - Conference contribution

AN - SCOPUS:85049371900

SN - 9783319930398

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 64

EP - 76

BT - Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings

A2 - Webb, Geoffrey I.

A2 - Phung, Dinh

A2 - Ganji, Mohadeseh

A2 - Rashidi, Lida

A2 - Tseng, Vincent S.

A2 - Ho, Bao

PB - Springer Verlag

T2 - 22nd Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2018

Y2 - 3 June 2018 through 6 June 2018

ER -

Montiel J, Read J, Bifet A, Abdessalem T. Scalable model-based cascaded imputation of missing data. In Webb GI, Phung D, Ganji M, Rashidi L, Tseng VS, Ho B, editors, Advances in Knowledge Discovery and Data Mining - 22nd Pacific-Asia Conference, PAKDD 2018, Proceedings. Springer Verlag. 2018. p. 64-76. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-93040-4_6

Scalable model-based cascaded imputation of missing data

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this