Search

Mapping deprivation in urban areas is important, for example for identifying areas of greatest need and planning interventions. Traditional ways of obtaining deprivation estimates are based on either census or household survey data, which in many areas is unavailable or difficult to collect. However, there has been a huge rise in the amount of new, non-traditional forms of data, such as satellite imagery and cell-phone call-record data, which may contain information useful for identifying deprivation. We use Angle-Based Joint and Individual Variation Explained (AJIVE) to jointly model satellite imagery data, cell-phone data, and survey data for the city of Dar es Salaam, Tanzania. We first identify interpretable low-dimensional structure from the imagery and cell-phone data, and find that we can use these to identify deprivation. We then consider what is gained from further incorporating the more traditional and costly survey data. We also introduce a scalar measure of deprivation as a response variable to be predicted, and consider various approaches to multiview regression, including using AJIVE scores as predictors., Comment: 34 pages, 25 figures

When mathematical biology models are used to make quantitative predictions for clinical or industrial use, it is important that these predictions come with a reliable estimate of their accuracy (uncertainty quantification). Because models of complex biological systems are always large simplifications, model discrepancy arises - where a mathematical model fails to recapitulate the true data generating process. This presents a particular challenge for making accurate predictions, and especially for making accurate estimates of uncertainty in these predictions. Experimentalists and modellers must choose which experimental procedures (protocols) are used to produce data to train their models. We propose to characterise uncertainty owing to model discrepancy with an ensemble of parameter sets, each of which results from training to data from a different protocol. The variability in predictions from this ensemble provides an empirical estimate of predictive uncertainty owing to model discrepancy, even for unseen protocols. We use the example of electrophysiology experiments, which are used to investigate the kinetics of the hERG potassium ion channel. Here, 'information-rich' protocols allow mathematical models to be trained using numerous short experiments performed on the same cell. Typically, assuming independent observational errors and training a model to an individual experiment results in parameter estimates with very little dependence on observational noise. Moreover, parameter sets arising from the same model applied to different experiments often conflict - indicative of model discrepancy. Our methods will help select more suitable mathematical models of hERG for future studies, and will be widely applicable to a range of biological modelling problems., Comment: Final published version with a typographical error in Table 1 (the value of q_6) corrected

Network data are becoming increasingly available, and so there is a need to develop suitable methodology for statistical analysis. Networks can be represented as graph Laplacian matrices, which are a type of manifold-valued data. Our main objective is to estimate a regression curve from a sample of graph Laplacian matrices conditional on a set of Euclidean covariates, for example in dynamic networks where the covariate is time. We develop an adapted Nadaraya-Watson estimator which has uniform weak consistency for estimation using Euclidean and power Euclidean metrics. We apply the methodology to the Enron email corpus to model smooth trends in monthly networks and highlight anomalous networks. Another motivating application is given in corpus linguistics, which explores trends in an author's writing style over time based on word co-occurrence networks., Comment: 15 pages, 4 figures

Word embeddings are commonly obtained as optimizers of a criterion function $f$ of a text corpus, but assessed on word-task performance using a different evaluation function $g$ of the test data. We contend that a possible source of disparity in performance on tasks is the incompatibility between classes of transformations that leave $f$ and $g$ invariant. In particular, word embeddings defined by $f$ are not unique; they are defined only up to a class of transformations to which $f$ is invariant, and this class is larger than the class to which $g$ is invariant. One implication of this is that the apparent superiority of one word embedding over another, as measured by word task performance, may largely be a consequence of the arbitrary elements selected from the respective solution sets. We provide a formal treatment of the above identifiability issue, present some numerical examples, and discuss possible resolutions., Comment: NIPS 2019

Networks arise in many applications, such as in the analysis of text documents, social interactions and brain activity. We develop a general framework for extrinsic statistical analysis of samples of networks, motivated by networks representing text documents in corpus linguistics. We identify networks with their graph Laplacian matrices, for which we define metrics, embeddings, tangent spaces, and a projection from Euclidean space to the space of graph Laplacians. This framework provides a way of computing means, performing principal component analysis and regression, and carrying out hypothesis tests, such as for testing for equality of means between two samples of networks. We apply the methodology to the set of novels by Jane Austen and Charles Dickens., Comment: 29 pages, 10 figures

A major goal in paleoclimate science is to reconstruct historical climates using proxies for climate variables such as those observed in sediment cores, and in the process learn about climate dynamics. This is hampered by uncertainties in how sediment core depths relate to ages, how proxy quantities relate to climate variables, how climate models are specified, and the values of parameters in climate models. Quantifying these uncertainties is key in drawing well founded conclusions. Analyses are often performed in separate stages with, for example, a sediment core's depth-age relation being estimated as stage one, then fed as an input to calibrate climate models as stage two. Here, we show that such "multi-stage" approaches can lead to misleading conclusions. We develop a joint inferential approach for climate reconstruction, model calibration, and age model estimation. We focus on the glacial-interglacial cycle over the past 780 kyr, analysing two sediment cores that span this range. Our age estimates are largely in agreement with previous studies, but provides the full joint specification of all uncertainties, estimation of model parameters, and the model evidence. By sampling plausible chronologies from the posterior distribution, we demonstrate that downstream scientific conclusions can differ greatly both between different sampled chronologies, and in comparison with conclusions obtained in the complete joint inferential analysis. We conclude that multi-stage analyses are insufficient when dealing with uncertainty, and that to draw sound conclusions the full joint inferential analysis must be performed.

We propose a two-parameter bounded probability distribution called the extended power distribution. This distribution on $(0, 1)$ is similar to the beta distribution, however there are some advantages which we explore. We define the moments and quantiles of this distribution and show that it is possible to give an $r$-parameter extension of this distribution ($r>2$). We also consider its complementary distribution and show that it has some flexibility advantages over the Kumaraswamy and beta distributions. This distribution can be used as an alternative to the Kumaraswamy distribution since it has a closed form for its cumulative function. However, it can be fitted to data where there are some samples that are exactly equal to 1, unlike the Kumaraswamy and beta distributions which cannot be fitted to such data or may require some censoring. Applications considered show the extended power distribution performs favourably against the Kumaraswamy distribution in most cases., Comment: 22 pages, 19 figures, 5 tables

Data of the form of event times arise in various applications. A simple model for such data is a non-homogeneous Poisson process (NHPP) which is specified by a rate function that depends on time. We consider the problem of having access to multiple independent observations of event time data, observed on a common interval, from which we wish to classify or cluster the observations according to their rate functions. Each rate function is unknown but assumed to belong to a finite number of rate functions each defining a distinct class. We model the rate functions using a spline basis expansion, the coefficients of which need to be estimated from data. The classification approach consists of using training data for which the class membership is known, to calculate maximum likelihood estimates of the coefficients for each group, then assigning test observations to a group by a maximum likelihood criterion. For clustering, by analogy to the Gaussian mixture model approach for Euclidean data, we consider mixtures of NHPP and use the expectation-maximisation algorithm to estimate the coefficients of the rate functions for the component models and group membership probabilities for each observation. The classification and clustering approaches perform well on both synthetic and real-world data sets. Code associated with this paper is available at https://github.com/duncan-barrack/NHPP ., Comment: cleaned up figures and text

In the context of data that lie on the simplex, we investigate use of empirical and exponential empirical likelihood, and Hotelling and James statistics, to test the null hypothesis of equal population means based on two independent samples. We perform an extensive numerical study using data simulated from various distributions on the simplex. The results, taken together with practical considerations regarding implementation, support the use of bootstrap-calibrated James statistic., Comment: This is a preprint of the article to be published by Taylor & Francis Group in Journal of Statistical Computation and Simulation

A prevailing viewpoint in palaeoclimate science is that a single palaeoclimate record contains insufficient information to discriminate between most competing explanatory models. Results we present here suggest the contrary. Using SMC^2 combined with novel Brownian bridge type proposals for the state trajectories, we show that even with relatively short time series it is possible to estimate Bayes factors to sufficient accuracy to be able to select between competing models. The results show that Monte Carlo methodology and computer power have now advanced to the point where a full Bayesian analysis for a wide class of conceptual climate models is now possible. The results also highlight a problem with estimating the chronology of the climate record prior to further statistical analysis, a practice which is common in palaeoclimate science. Using two datasets based on the same record but with different estimated chronologies results in conflicting conclusions about the importance of the orbital forcing on the glacial cycle, and about the internal dynamics generating the glacial cycle, even though the difference between the two estimated chronologies is consistent with dating uncertainty. This highlights a need for chronology estimation and other inferential questions to be addressed in a joint statistical procedure.

The characterisation of time-series data via their most salient features is extremely important in a range of machine learning task, not least of all with regards to classification and clustering. While there exist many feature extraction techniques suitable for non-intermittent time-series data, these approaches are not always appropriate for intermittent time-series data, where intermittency is characterized by constant values for large periods of time punctuated by sharp and transient increases or decreases in value. Motivated by this, we present aggregation, mode decomposition and projection (AMP) a feature extraction technique particularly suited to intermittent time-series data which contain time-frequency patterns. For our method all individual time-series within a set are combined to form a non-intermittent aggregate. This is decomposed into a set of components which represent the intrinsic time-frequency signals within the data set. Individual time-series can then be fit to these components to obtain a set of numerical features that represent their intrinsic time-frequency patterns. To demonstrate the effectiveness of AMP, we evaluate against the real word task of clustering intermittent time-series data. Using synthetically generated data we show that a clustering approach which uses the features derived from AMP significantly outperforms traditional clustering methods. Our technique is further exemplified on a real world data set where AMP can be used to discover groupings of individuals which correspond to real world sub-populations., Comment: Paper accepted in workshop on mining and learning from time series (MiLeTS) to be held in conjunction with KDD 2015 August 10- 13 2015, Sydney, Australia

In compositional data analysis an observation is a vector containing non-negative values, only the relative sizes of which are considered to be of interest. Without loss of generality, a compositional vector can be taken to be a vector of proportions that sum to one. Data of this type arise in many areas including geology, archaeology, biology, economics and political science. In this paper we investigate methods for classification of compositional data. Our approach centres on the idea of using the $\alpha$-transformation to transform the data and then to classify the transformed data via regularised discriminant analysis and the k-nearest neighbours algorithm. Using the $\alpha$-transformation generalises two rival approaches in compositional data analysis, one (when $\alpha=1$) that treats the data as though they were Euclidean, ignoring the compositional constraint, and another (when $\alpha=0$) that employs Aitchison's centred log-ratio transformation. A numerical study with several real datasets shows that whether using $\alpha=1$ or $\alpha=0$ gives better classification performance depends on the dataset, and moreover that using an intermediate value of $\alpha$ can sometimes give better performance than using either 1 or 0., Comment: This is a 17-page preprint and has been accepted for publication at the Journal of Classification

Many modern statistical applications involve inference for complicated stochastic models for which the likelihood function is difficult or even impossible to calculate, and hence conventional likelihood-based inferential echniques cannot be used. In such settings, Bayesian inference can be performed using Approximate Bayesian Computation (ABC). However, in spite of many recent developments to ABC methodology, in many applications the computational cost of ABC necessitates the choice of summary statistics and tolerances that can potentially severely bias the estimate of the posterior. We propose a new "piecewise" ABC approach suitable for discretely observed Markov models that involves writing the posterior density of the parameters as a product of factors, each a function of only a subset of the data, and then using ABC within each factor. The approach has the advantage of side-stepping the need to choose a summary statistic and it enables a stringent tolerance to be set, making the posterior "less approximate". We investigate two methods for estimating the posterior density based on ABC samples for each of the factors: the first is to use a Gaussian approximation for each factor, and the second is to use a kernel density estimate. Both methods have their merits. The Gaussian approximation is simple, fast, and probably adequate for many applications. On the other hand, using instead a kernel density estimate has the benefit of consistently estimating the true ABC posterior as the number of ABC samples tends to infinity. We illustrate the piecewise ABC approach for three examples; in each case, the approach enables "exact matching" between simulations and data and offers fast and accurate inference.

Compositional data analysis is carried out either by neglecting the compositional constraint and applying standard multivariate data analysis, or by transforming the data using the logs of the ratios of the components. In this work we examine a more general transformation which includes both approaches as special cases. It is a power transformation and involves a single parameter, {\alpha}. The transformation has two equivalent versions. The first is the stay-in-the-simplex version, which is the power transformation as defined by Aitchison in 1986. The second version, which is a linear transformation of the power transformation, is a Box-Cox type transformation. We discuss a parametric way of estimating the value of {\alpha}, which is maximization of its profile likelihood (assuming multivariate normality of the transformed data) and the equivalence between the two versions is exhibited. Other ways include maximization of the correct classification probability in discriminant analysis and maximization of the pseudo R-squared (as defined by Aitchison in 1986) in linear regression. We examine the relationship between the {\alpha}-transformation, the raw data approach and the isometric log-ratio transformation. Furthermore, we also define a suitable family of metrics corresponding to the family of {\alpha}-transformation and consider the corresponding family of Frechet means., Comment: Published in the proceddings of the 4th international workshop on Compositional Data Analysis. http://congress.cimne.com/codawork11/frontal/default.asp