arXiv subjects

Piotr Kokoszka

Publications and source records attributed to Piotr Kokoszka.

29 records · Page 2Linked to original sources

arXiv2015

Testing separability of space--time functional processes

We present a new methodology and accompanying theory to test for separability of spatio-temporal functional data. In spatio-temporal statistics, separability is a common simplifying assumption concerning the covariance structure which, if true, can greatly increase estimation accuracy and inferential power. While our focus is on testing for the separation of space and time in spatio-temporal data, our methods can be applied to any area where separability is useful, including biomedical imaging. We present three tests, one being a functional extension of the Monte Carlo likelihood method of Mitchell et. al. (2005), while the other two are based on quadratic forms. Our tests are based on asymptotic distributions of maximum likelihood estimators, and do not require Monte Carlo or bootstrap replications. The specification of the joint asymptotic distribution of these estimators is the main theoretical contribution of this paper. It can be used to derive many other tests. The main methodological finding is that one of the quadratic form methods, which we call a norm approach, emerges as a clear winner in terms of finite sample performance in nearly every setting we considered. The norm approach focuses directly on the Frobenius distance between the spatio-temporal covariance function and its separable approximation. We demonstrate the efficacy of our methods via simulations and an application to Irish wind data.

stat.ME↗

arXiv2013

Consistency of the mean and the principal components of spatially distributed functional data

This paper develops a framework for the estimation of the functional mean and the functional principal components when the functions form a random field. More specifically, the data we study consist of curves $X(\mathbf{s}_k;t),t\in[0,T]$, observed at spatial points $\mathbf{s}_1,\mathbf{s}_2,\ldots,\mathbf{s}_N$. We establish conditions for the sample average (in space) of the $X(\mathbf{s}_k)$ to be a consistent estimator of the population mean function, and for the usual empirical covariance operator to be a consistent estimator of the population covariance operator. These conditions involve an interplay of the assumptions on an appropriately defined dependence between the functions $X(\mathbf{s}_k)$ and the assumptions on the spatial distribution of the points $\mathbf{s}_k$. The rates of convergence may be the same as for i.i.d. functional samples, but generally depend on the strength of dependence and appropriately quantified distances between the points $\mathbf{s}_k$. We also formulate conditions for the lack of consistency.

math.ST↗

arXiv2013

Functional Data Analysis with Increasing Number of Projections

Functional principal components (FPC's) provide the most important and most extensively used tool for dimension reduction and inference for functional data. The selection of the number, d, of the FPC's to be used in a specific procedure has attracted a fair amount of attention, and a number of reasonably effective approaches exist. Intuitively, they assume that the functional data can be sufficiently well approximated by a projection onto a finite-dimensional subspace, and the error resulting from such an approximation does not impact the conclusions. This has been shown to be a very effective approach, but it is desirable to understand the behavior of many inferential procedures by considering the projections on subspaces spanned by an increasing number of the FPC's. Such an approach reflects more fully the infinite-dimensional nature of functional data, and allows to derive procedures which are fairly insensitive to the selection of d. This is accomplished by considering limits as d tends to infinity with the sample size. We propose a specific framework in which we let d tend to infinity by deriving a normal approximation for the two-parameter partial sum process of the scores ξ_{i,j} of the i-th function with respect to the j-th FPC. Our approximation can be used to derive statistics that use segments of observations and segments of the FPC's. We apply our general results to derive two inferential procedures for the mean function: a change-point test and a two-sample test. In addition to the asymptotic theory, the tests are assessed through a small simulation study and a data example.

math.ST↗

arXiv2012

Estimation and testing for spatially indexed curves with application to ionospheric and magnetic field trends

We develop methodology for the estimation of the functional mean and the functional principal components when the functions form a spatial process. The data consist of curves $X(\mathbf{s}_k;t),t\in[0,T],$ observed at spatial locations $\mathbf{s}_1,\mathbf{s}_2,...,\mathbf{s}_N$. We propose several methods, and evaluate them by means of a simulation study. Next, we develop a significance test for the correlation of two such functional spatial fields. After validating the finite sample performance of this test by means of a simulation study, we apply it to determine if there is correlation between long-term trends in the so-called critical ionospheric frequency and decadal changes in the direction of the internal magnetic field of the Earth. The test provides conclusive evidence for correlation, thus solving a long-standing space physics conjecture. This conclusion is not apparent if the spatial dependence of the curves is neglected.

stat.AP↗

arXiv2011

Estimation of the mean of functional time series and a two sample problem

This paper is concerned with inference based on the mean function of a functional time series, which is defined as a collection of curves obtained by splitting a continuous time record, e.g. into daily or annual curves. We develop a normal approximation for the functional sample mean, and then focus on the estimation of the asymptotic variance kernel. Using these results, we develop and asymptotically justify a testing procedure for the equality of means in two functional samples exhibiting temporal dependence. Evaluated by means of a simulations study and application to real data sets, this two sample procedure enjoys good size and power in finite samples. We provide the details of its numerical implementation.

math.ST↗

arXiv2011

Testing the Equality of Covariance Operators in Functional Samples

We propose a robust test for the equality of the covariance structures in two functional samples. The test statistic has a chi-square asymptotic distribution with a known number of degrees of freedom, which depends on the level of dimension reduction needed to represent the data. Detailed analysis of the asymptotic properties is developed. Finite sample performance is examined by a simulation study and an application to egg-laying curves of fruit flies.

stat.ME↗

arXiv2010

Weakly dependent functional data

Functional data often arise from measurements on fine time grids and are obtained by separating an almost continuous time record into natural consecutive intervals, for example, days. The functions thus obtained form a functional time series, and the central issue in the analysis of such data consists in taking into account the temporal dependence of these functional observations. Examples include daily curves of financial transaction data and daily patterns of geophysical and environmental data. For scalar and vector valued stochastic processes, a large number of dependence notions have been proposed, mostly involving mixing type distances between $σ$-algebras. In time series analysis, measures of dependence based on moments have proven most useful (autocovariances and cumulants). We introduce a moment-based notion of dependence for functional time series which involves $m$-dependence. We show that it is applicable to linear as well as nonlinear functional time series. Then we investigate the impact of dependence thus quantified on several important statistical procedures for functional data. We study the estimation of the functional principal components, the long-run covariance matrix, change point detection and the functional linear model. We explain when temporal dependence affects the results obtained for i.i.d. functional observations and when these results are robust to weak dependence.

math.ST↗

arXiv2008

Testing for changes in polynomial regression

We consider a nonlinear polynomial regression model in which we wish to test the null hypothesis of structural stability in the regression parameters against the alternative of a break at an unknown time. We derive the extreme value distribution of a maximum-type test statistic which is asymptotically equivalent to the maximally selected likelihood ratio. The resulting test is easy to apply and has good size and power, even in small samples.

math.ST↗

arXiv2008

Sample autocovariances of long-memory time series

We find the asymptotic distribution of the sample autocovariances of long-memory processes in cases of finite and infinite fourth moment. Depending on the interplay of assumptions on moments and the intensity of dependence, there are three types of convergence rates and limit distributions. In particular, a normal approximation with the standard rate does not always hold in practically relevant cases.

math.ST↗

arXiv2006

On discriminating between long-range dependence and changes in mean

We develop a testing procedure for distinguishing between a long-range dependent time series and a weakly dependent time series with change-points in the mean. In the simplest case, under the null hypothesis the time series is weakly dependent with one change in mean at an unknown point, and under the alternative it is long-range dependent. We compute the CUSUM statistic $T_n$, which allows us to construct an estimator $\hat{k}$ of a change-point. We then compute the statistic $T_{n,1}$ based on the observations up to time $\hat{k}$ and the statistic $T_{n,2}$ based on the observations after time $\hat{k}$. The statistic $M_n=\max[T_{n,1},T_{n,2}]$ converges to a well-known distribution under the null, but diverges to infinity if the observations exhibit long-range dependence. The theory is illustrated by examples and an application to the returns of the Dow Jones index.

math.ST↗

arXiv2005

Near-integrated GARCH sequences

Motivated by regularities observed in time series of returns on speculative assets, we develop an asymptotic theory of GARCH(1,1) processes {y_k} defined by the equations y_k=σ_kε_k, σ_k^2=ω+αy_{k-1}^2+βσ_{k-1}^2 for which the sum α+βapproaches unity as the number of available observations tends to infinity. We call such sequences near-integrated. We show that the asymptotic behavior of near-integrated GARCH(1,1) processes critically depends on the sign of γ:=α+β-1. We find assumptions under which the solutions exhibit increasing oscillations and show that these oscillations grow approximately like a power function if γ\leq 0 and exponentially if γ>0. We establish an additive representation for the near-integrated GARCH(1,1) processes which is more convenient to use than the traditional multiplicative Volterra series expansion.

math.PR↗