What is the method for evaluating missing data?

The authors recommend its use if, as well as excellent classification accuracy, it is desirable that the end-user understand the methodology employed.

Open AccessJournal Article10.1016/J.PATREC.2008.01.010

Good methods for coping with missing data in decision trees

Bhekisipho Twala, +2 more

- 01 May 2008

- Pattern Recognition Letters

- Vol. 29, Iss: 7, pp 950-956

170

TL;DR: It is shown through a substantial data-based study of classification accuracy that MIA exhibits consistently good performance across a broad range of data types and of sources and amounts of missingness.

About: This article is published in Pattern Recognition Letters. The article was published on 01 May 2008. and is currently open access. The article focuses on the topics: Missing data & Categorical variable.

AI Agents for this Paper

Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps

Most frequently asked questions

1. What are the contributions mentioned in the paper "Good methods for coping with missing data in decision trees" ?

The authors propose a simple and effective method for dealing with missing data in decision trees used for classification.. The authors show through a substantial data-based study of classification accuracy that MIA exhibits consistently good performance across a broad range of data types and of sources and amounts of missingness.

2. What are the missingness mechanisms used in the study?

These were: deletion of instances with any missing data; Shapiro's decision tree single imputation technique (Quinlan, 1993); maximum likelihood imputation of data for both continuous (assuming multivariate normality) and categorical data via the EM algorithm as developed by Schafer (1997), and considered in both single and (five replicate) multiple imputation (EMMI) forms; mean or mode single imputation; fractional cases (FC; Cestnik et al., 1987, Quinlan, 1993); and surrogate variable splitting (Breiman et al., 1984, Therneau and Atkinson, 1997).

3. What are the missing data mechanisms used in the study?

Three missing data mechanisms were employed, coming under the headings (Little and Rubin, 2002) of: missing completely at random (MCAR); missing at random (MAR), under which the probability of being missing depends on the value of another, non-missing, attribute; and informative missingness (IM), under which the probability of being missing depends on the actual (but in non-simulation practice unobserved) value of the attribute itself.

4. What are the three missingness mechanisms used to cope with missing data?

All are complete datasets into which missingness is artificially introduced at rates of 15%, 30% and 50% into either the single attribute which is most highly correlated with class or else evenly distributed across all attributes.

Fig. 1. Excess error rates attained by MIA, EMMI and FC, averaged over the 21 datasets and plotted against percentages of missing values. The six frames correspond to MCARuniva, MCARunifo, MARuniva, MARunifo, IMuniva and IMunifo missingness mechanism/distribution combinations.

Fig. 2. Excess error rates attained by MIA, EMMI and FC for the `kr-vs-kp' dataset, plotted against percentages of missing values. The six frames correspond to MCARuniva, MCARunifo, MARuniva, MARunifo, IMuniva and IMunifo missingness mechanism/distribution combinations.

Fig. 3. Excess error rates attained by EMIMIA and REMIMIA as well as MIA and EMMI, averaged over the 21 datasets and plotted against percentages of missing values. The six frames correspond to MCARuniva, MCARunifo, MARuniva, MARunifo, IMuniva and IMunifo missingness mechanism/distribution combinations.

Citations

Journal Article•10.1007/S10462-011-9272-4

Decision trees: a recent overview

Sotiris Kotsiantis

- 01 Apr 2013

- Artificial Intelligence Review

TL;DR: Basic decision tree issues and current research points are described, guiding the researcher in interesting research directions and suggesting possible bias combinations that have yet to be explored.

...read moreread less

1.3K

•Journal Article•10.1002/SAM.11348

Random Forest Missing Data Algorithms.

Fei Tang, +1 more

- 13 Jun 2017

- Statistical Analysis and Data Mining

TL;DR: RF imputation is revealed to be generally robust with performance improving with increasing correlation, and performance was good under moderate to high missingness, and even when data was missing not at random.

...read moreread less

604

Journal Article•10.1007/S10462-019-09709-4

Missing value imputation: a review and analysis of the literature (2006–2017)

Wei Chao Lin, +2 more

- 01 Feb 2020

- Artificial Intelligence Review

TL;DR: This paper aims at reviewing and analyzing related studies carried out in recent decades, from the experimental design perspective, and identifying limitations in the existing body of literature based upon which some directions for future research can be gleaned.

...read moreread less

473

•Journal Article•10.18637/JSS.V070.I04

bartMachine: Machine Learning with Bayesian Additive Regression Trees

Adam Kapelner, +1 more

- 04 Apr 2016

- Journal of Statistical Software

TL;DR: In this paper, the authors present a new package in R implementing Bayesian additive regression trees (BART), which introduces many new features for data analysis using BART such as variable selection, interaction detection, model diagnostic plots, incorporation of missing data and the ability to save trees for future prediction.

...read moreread less

323

•Posted Content

bartMachine: Machine Learning with Bayesian Additive Regression Trees

Adam Kapelner, +1 more

- 08 Dec 2013

- arXiv: Machine Learning

TL;DR: The package introduces many new features for data analysis using BART such as variable selection, interaction detection, model diagnostic plots, incorporation of missing data and the ability to save trees for future prediction.

...read moreread less

167

References

•Book

C4.5: Programs for Machine Learning

J. Ross Quinlan

- 15 Oct 1992

TL;DR: A complete guide to the C4.5 system as implemented in C for the UNIX environment, which starts from simple core learning methods and shows how they can be elaborated and extended to deal with typical problems such as missing data and over hitting.

...read moreread less

27.2K

•Book

Classification and regression trees

Leo Breiman

- 01 Jan 1983

TL;DR: The methodology used to construct tree structured rules is the focus of a monograph as mentioned in this paper, covering the use of trees as a data analysis method, and in a more mathematical framework, proving some of their fundamental properties.

...read moreread less

22.7K

Journal Article•10.2307/2288003

Classification and Regression Trees.

John Van Ryzin, +4 more

- 01 Mar 1986

- Journal of the American Statistical Asso...

21.8K

•Book

The Elements of Statistical Learning: Data Mining, Inference, and Prediction

Trevor Hastie, +2 more

- 28 Jul 2013

TL;DR: In this paper, the authors describe the important ideas in these areas in a common conceptual framework, and the emphasis is on concepts rather than mathematics, with a liberal use of color graphics.

...read moreread less

21.3K

•Journal Article•10.1023/A:1022643204877

Induction of Decision Trees

J. R. Quinlan

- 25 Mar 1986

- Machine Learning

TL;DR: In this paper, an approach to synthesizing decision trees that has been used in a variety of systems, and it describes one such system, ID3, in detail, is described, and a reported shortcoming of the basic algorithm is discussed.

...read moreread less

18.8K

Good methods for coping with missing data in decision trees

Chat with Paper

AI Agents for this Paper

Most frequently asked questions

1. What are the contributions mentioned in the paper "Good methods for coping with missing data in decision trees" ?

2. What are the missingness mechanisms used in the study?

3. What are the missing data mechanisms used in the study?

4. What are the three missingness mechanisms used to cope with missing data?

5. What is the method for evaluating missing data?

Figures

Citations

Decision trees: a recent overview

Random Forest Missing Data Algorithms.

Missing value imputation: a review and analysis of the literature (2006–2017)

bartMachine: Machine Learning with Bayesian Additive Regression Trees

bartMachine: Machine Learning with Bayesian Additive Regression Trees

References

C4.5: Programs for Machine Learning

Classification and regression trees

Classification and Regression Trees.

The Elements of Statistical Learning: Data Mining, Inference, and Prediction

Induction of Decision Trees

Related Papers (5)

Statistical Analysis with Missing Data

Random Forests

Inference and missing data

C4.5: Programs for Machine Learning

Classification and Regression Trees.