A comparison of machine learning methods for classification using simulation with multiple real data examples from mental health studies
Open Access
- 30 September 2016
- journal article
- research article
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 25 (5) , 1804-1823
Abstract
Background: Recent literature on the comparison of machine learning methods has raised questions about the neutrality, unbiasedness and utility of many comparative studies. Reporting of results on favourable datasets and sampling error in the estimated performance measures based on single samples are thought to be the major sources of bias in such comparisons. Better performance in one or a few instances does not necessarily imply so on an average or on a population level and simulation studies may be a better alternative for objectively comparing the performances of machine learning algorithms. Methods: We compare the classification performance of a number of important and widely used machine learning algorithms, namely the Random Forests (RF), Support Vector Machines (SVM), Linear Discriminant Analysis (LDA) and k-Nearest Neighbour (kNN). Using massively parallel processing on high-performance supercomputers, we compare the generalisation errors at various combinations of levels of several factors: number of features, training sample size, biological variation, experimental variation, effect size, replication and correlation between features. Results: For smaller number of correlated features, number of features not exceeding approximately half the sample size, LDA was found to be the method of choice in terms of average generalisation errors as well as stability (precision) of error estimates. SVM (with RBF kernel) outperforms LDA as well as RF and kNN by a clear margin as the feature set gets larger provided the sample size is not too small (at least 20). The performance of kNN also improves as the number of features grows and outplays that of LDA and RF unless the data variability is too high and/or effect sizes are too small. RF was found to outperform only kNN in some instances where the data are more variable and have smaller effect sizes, in which cases it also provide more stable error estimates than kNN and LDA. Applications to a number of real datasets supported the findings from the simulation study.Keywords
This publication has 33 references indexed in Scilit:
- Regional Magnetic Resonance Imaging Measures for Multivariate Analysis in Alzheimer’s Disease and Mild Cognitive ImpairmentBrain Topography, 2012
- Multiple-rule bias in the comparison of classification rulesBioinformatics, 2011
- Electrophysiological evidence for abnormal preparatory states and inhibitory processing in adult ADHDBehavioral and Brain Functions, 2010
- MRI Measures of Alzheimer's Disease and the AddNeuroMed StudyAnnals of the New York Academy of Sciences, 2009
- A comprehensive comparison of random forests and support vector machines for microarray-based cancer classificationBMC Bioinformatics, 2008
- A comparative study of different machine learning methods on microarray gene expression dataBMC Genomics, 2008
- Linear Models and Empirical Bayes Methods for Assessing Differential Expression in Microarray ExperimentsStatistical Applications in Genetics and Molecular Biology, 2004
- The Lack of A Priori Distinctions Between Learning AlgorithmsNeural Computation, 1996
- Support-vector networksMachine Learning, 1995
- Estimating the Error Rate of a Prediction Rule: Improvement on Cross-ValidationJournal of the American Statistical Association, 1983