We introduce a very general method for high dimensional classification, based on careful combination of the results of applying an arbitrary base classifier to random projections of the feature vectors into a lower dimensional space. In one special case that we study in detail, the random projections are divided into disjoint groups, and within each group we select the projection yielding the smallest estimate of the test error. Our random-projection ensemble classifier then aggregates the results of applying the base classifier on the selected projections, with a data-driven voting threshold to determine the final assignment. Our theoretical results elucidate the effect on performance of increasing the number of projections. Moreover, under a boundary condition that is implied by the sufficient dimension reduction assumption, we show that the test excess risk of the random-projection ensemble classifier can be controlled by terms that do not depend on the original data dimension and a term that becomes negligible as the number of projections increases. The classifier is also compared empirically with several other popular high dimensional classifiers via an extensive simulation study, which reveals its excellent finite sample performance.

No abstract

No abstract

We study the effect of imperfect training data labels on the performance of classification methods. In a general setting, where the probability that an observation in the training dataset is mislabelled may depend on both the feature vector and the true label, we bound the excess risk of an arbitrary classifier trained with imperfect labels in terms of its excess risk for predicting a noisy label. This reveals conditions under which a classifier trained with imperfect labels remains consistent for classifying uncorrupted test data points. Furthermore, under stronger conditions, we derive detailed asymptotic properties for the popular k-nearest neighbour (knn), support vector machine (SVM) and linear discriminant analysis (LDA) classifiers. One consequence of these results is that the knn and SVM classifiers are robust to imperfect training labels, in the sense that the rate of convergence of the excess risks of these classifiers remains unchanged; in fact, our theoretical and empirical results even show that in some cases, imperfect labels may improve the performance of these methods. On the other hand, the LDA classifier is shown to be typically inconsistent in the presence of label noise unless the prior probabilities of each class are equal. Our theoretical results are supported by a simulation study.

We derive a new asymptotic expansion for the global excess risk of a local-k-nearest neighbour classifier, where the choice of k may depend upon the test point. This expansion elucidates conditions under which the dominant contribution to the excess risk comes from the decision boundary of the optimal Bayes classifier, but we also show that if these conditions are not satisfied, then the dominant contribution may arise from the tails of the marginal distribution of the features. Moreover, we prove that, provided the d-dimensional marginal distribution of the features has a finite ρth moment for some ρ > 4 (as well as other regularity conditions), a local choice of k can yield a rate of convergence of the excess risk of O(n −4/(d+4) ), where n is the sample size, whereas for the standard k-nearest neighbour classifier, our theory would require d ≥ 5 and ρ > 4d/(d − 4) finite moments to achieve this rate. These results motivate a new k-nearest neighbour classifier for semi-supervised learning problems, where the unlabelled data are used to obtain an estimate of the marginal feature density, and fewer neighbours are used for classification when this density estimate is small. Our worst-case rates are complemented by a minimax lower bound, which reveals that the local, semi-supervised k-nearest neighbour classifier attains the minimax optimal rate over our classes for the excess risk, up to a subpolynomial factor in n. These theoretical improvements over the standard k-nearest neighbour classifier are also illustrated through a simulation study.

No abstract

We derive a new asymptotic expansion for the global excess risk of a local-k-nearest neighbour classifier, where the choice of k may depend upon the test point. This expansion elucidates conditions under which the dominant contribution to the excess risk comes from the decision boundary of the optimal Bayes classifier, but we also show that if these conditions are not satisfied, then the dominant contribution may arise from the tails of the marginal distribution of the features. Moreover, we prove that, provided the d-dimensional marginal distribution of the features has a finite ρth moment for some ρ > 4 (as well as other regularity conditions), a local choice of k can yield a rate of convergence of the excess risk of O(n −4/(d+4) ), where n is the sample size, whereas for the standard k-nearest neighbour classifier, our theory would require d ≥ 5 and ρ > 4d/(d − 4) finite moments to achieve this rate. These results motivate a new k-nearest neighbour classifier for semi-supervised learning problems, where the unlabelled data are used to obtain an estimate of the marginal feature density, and fewer neighbours are used for classification when this density estimate is small. Our worst-case rates are complemented by a minimax lower bound, which reveals that the local, semi-supervised k-nearest neighbour classifier attains the minimax optimal rate over our classes for the excess risk, up to a subpolynomial factor in n. These theoretical improvements over the standard k-nearest neighbour classifier are also illustrated through a simulation study.

Type 2 diabetes mellitus (T2D) is one of the most prevalent diseases in the world and presents a major health and economic burden, a notable proportion of which could be alleviated with improved early prediction and intervention. While standard risk factors including age, obesity, and hypertension have shown good predictive performance, we show that the use of CpG DNA methylation information leads to a significant improvement in the prediction of 10-year T2D incidence risk.Whilst previous studies have been largely constrained by linear assumptions and the use of CpGs one-at-the-time, we have adopted a more flexible approach based on a range of linear and tree-ensemble models for classification and time-to-event prediction. Using the Generation Scotland cohort (n=9,537) our best performing model (Area Under the Curve (AUC)=0.880, Precision Recall AUC (PRAUC)=0.539, McFadden’s R2=0.316) used a LASSO Cox proportional-hazards predictor and showed notable improvement in onset prediction, above and beyond standard risk factors (AUC=0.860, PRAUC=0.444 R2=0.261). Replication of the main finding was observed in an external test dataset (the German-based KORA study, p=3.7×10−4). Tree-ensemble methods provided comparable performance and future improvements to these models are discussed.Finally, we introduce MethylPipeR, an R package with accompanying user interface, for systematic and reproducible development of complex trait and incident disease predictors. While MethylPipeR was applied to incident T2D prediction with DNA methylation in our experiments, the package is designed for generalised development of predictive models and is applicable to a wide range of omics data and target traits.