Harrison–Stevens Forecasting and the Multiprocess Dynamic Linear Model

This is an expository article. The Harrison–Stevens forecasting algorithm using the multiprocess dynamic linear model is a robust method for forecasting in a nonstationary time series. The purpose of this article is to help statisticians become familiar with the method.     

Monte Carlo Kalman filter and smoothing for multivariate discrete state space models

The author studies state space models for multivariate binomial time series, focussing on the development of the Kalman filter and smoothing for state variables. He proposes a Monte Carlo approach employing the latent variable representation which transplants the classical Kalman filter and smoothing developed for Gaussian state space models to discrete models and leads to a conceptually simple and computationally convenient approach. The method is illustrated through simulations and concrete examples.     

Missing data and forecasting in multivariate time series: An application of the common components dynamic linear model

Summary The paper deals with missing data and forecasting problems in multivariate time series making use of the Common Components Dynamic Linear Model (DLMCC), presented in Quintana (1985), and West and Harrison (1989). Some results are presented and discussed: exploiting the correlation between series, estimated by the DLMCC, the paper shows as it is possible to update state vector posterior distributions for the unobserved series. This is realized on the base of the updating of the observed series state vectors, for which the usual Kalman filter equations can be applied. An application concerning some Italian private consumption series provides an example of the model capabilities.     

Forecasting Multivariate Road Traffic Flows Using Bayesian Dynamic Graphical Models,Splines and Other Traffic Variables

Traffic flow data are routinely collected for many networks worldwide. These invariably large data sets can be used as part of a traffic management system, for which good traffic flow forecasting models are crucial. The linear multiregression dynamic model (LMDM) has been shown to be promising for forecasting flows, accommodating multivariate flow time series, while being a computationally simple model to use. While statistical flow forecasting models usually base their forecasts on flow data alone, data for other traffic variables are also routinely collected. This paper shows how cubic splines can be used to incorporate extra variables into the LMDM in order to enhance flow forecasts. Cubic splines are also introduced into the LMDM to parsimoniously accommodate the daily cycle exhibited by traffic flows. The proposed methodology allows the LMDM to provide more accurate forecasts when forecasting flows in a real high‐dimensional traffic data set. The resulting extended LMDM can deal with some important traffic modelling issues not usually considered in flow forecasting models. Additionally, the model can be implemented in a real‐time environment, a crucial requirement for traffic management systems designed to support decisions and actions to alleviate congestion and keep traffic flowing.     

Estimation and tests of hypotheses for the initial mean and covariance in the kalman filter model

Kalman filtering techniques are widely used by engineers to recursively estimate random signal parameters which are essentially coefficients in a large-scale time series regression model. These Bayesian estimators depend on the values assumed for the mean and covariance parameters associated with the initial state of the random signal. This paper considers a likelihood approach to estimation and tests of hypotheses involving the critical initial means and covariances. A computationally simple convergent iterative algorithm is used to generate estimators which depend only on standard Kalman filter outputs at each successive stage. Conditions are given under which the maximum likelihood estimators are consistent and asymptotically normal. The procedure is illustrated using a typical large-scale data set involving 10-dimensional signal vectors.     

Multivariate stochastic volatility with Bayesian dynamic linear models

This paper develops a Bayesian procedure for estimation and forecasting of the volatility of multivariate time series. The foundation of this work is the matrix-variate dynamic linear model, for the volatility of which we adopt a multiplicative stochastic evolution, using Wishart and singular multivariate beta distributions. A diagonal matrix of discount factors is employed in order to discount the variances element by element and therefore allowing a flexible and pragmatic variance modelling approach. Diagnostic tests and sequential model monitoring are discussed in some detail. The proposed estimation theory is applied to a four-dimensional time series, comprising spot prices of aluminium, copper, lead and zinc of the London metal exchange. The empirical findings suggest that the proposed Bayesian procedure can be effectively applied to financial data, overcoming many of the disadvantages of existing volatility models.     

Forecasting Contemporaneously Aggregated Vector ARMA Processes

Given a multiple time series that is generated by a multivariate ARMA process and assuming the objective is to forecast a weighted sum of the individual variables, then under a mean squared error measure of forecasting precision, it is preferable to forecast the disaggregated multiple time series and aggregate the forecasts, rather than forecast the aggregated series directly, if the involved processes are known. This result fails to hold if the processes used for forecasting are estimated from a given set of time series data. The implications of these results for empirical research are investigated using different sets of economic data.     

Reference Priors for Matrix-Variate Dynamic Linear Models

We develop reference analysis for matrix-variate dynamic models with unknown observation covariance matrices. Bayesian algorithms for forecasting, estimation, and filtering are derived. This work extends the existing theory of reference analysis for univariate dynamic linear models, and thus it proposes a solution to the specification of the prior distributions for a very wide class of time series models. Subclasses of our models include the widely used multivariate and matrix-variate regression models.     

Forecasting Economic Time Series With Structural and Box-Jenkins Models: A Case Study

The basic structural model is a univariate time series model consisting of a slowly changing trend component, a slowly changing seasonal component, and a random irregular component. It is part of a class of models that have a number of advantages over the seasonal ARIMA models adopted by Box and Jenkins (1976). This article reports the results of an exercise in which the basic structural model was estimated for six U.K. macroeconomic time series and the forecasting performance compared with that of ARIMA models previously fitted by Prothero and Wallis (1976).     

Dynamic structural models with covariates for short-term forecasting of time series with complex seasonal patterns

This work presents a framework of dynamic structural models with covariates for short-term forecasting of time series with complex seasonal patterns. The framework is based on the multiple sources of randomness formulation. A noise model is formulated to allow the incorporation of randomness into the seasonal component and to propagate this same randomness in the coefficients of the variant trigonometric terms over time. A unique, recursive and systematic computational procedure based on the maximum likelihood estimation under the hypothesis of Gaussian errors is introduced. The referred procedure combines the Kalman filter with recursive adjustment of the covariance matrices and the selection method of harmonics number in the trigonometric terms. A key feature of this method is that it allows estimating not only the states of the system but also allows obtaining the standard errors of the estimated parameters and the prediction intervals. In addition, this work also presents a non-parametric bootstrap approach to improve the forecasting method based on Kalman filter recursions. The proposed framework is empirically explored with two real time series.     

Short-term forecasting with a computationally efficient nonparametric transfer function model

In this paper a semi-parametric approach is developed to model non-linear relationships in time series data using polynomial splines. Polynomial splines require very little assumption about the functional form of the underlying relationship, so they are very flexible and can be used to model highly non-linear relationships. Polynomial splines are also computationally very efficient. The serial correlation in the data is accounted for by modelling the noise as an autoregressive integrated moving average (ARIMA) process, by doing so, the efficiency in nonparametric estimation is improved and correct inferences can be obtained. The explicit structure of the ARIMA model allows the correlation information to be used to improve forecasting performance. An algorithm is developed to automatically select and estimate the polynomial spline model and the ARIMA model through backfitting. This method is applied on a real-life data set to forecast hourly electricity usage. The non-linear effect of temperature on hourly electricity usage is allowed to be different at different hours of the day and days of the week. The forecasting performance of the developed method is evaluated in post-sample forecasting and compared with several well-accepted models. The results show the performance of the proposed model is comparable with a long short-term memory deep learning model.     

Multivariate temporal disaggregation with cross-sectional constraints

Multivariate temporal disaggregation deals with the historical reconstruction and nowcasting of economic variables subject to temporal and contemporaneous aggregation constraints. The problem involves a system of time series that are related not only by a dynamic model but also by accounting constraints. The paper introduces two fundamental (and realistic) models that implement the multivariate best linear unbiased estimation approach that has potential application to the temporal disaggregation of the national accounts series. The multivariate regression model with random walk disturbances is most suitable to deal with the chained linked volumes (as the nature of the national accounts time series suggests); however, in this case the accounting constraints are not binding and the discrepancy has to be modeled by either a trend-stationary or an integrated process. The tiny, compared with other driving disturbances, size of the discrepancy prevents maximum-likelihood estimation to be carried out, and the parameters have to be estimated separately. The multivariate disaggregation with integrated random walk disturbances is suitable for the national accounts aggregates expressed at current prices, in which case the accounting constraints are binding.     

On-line quantization in nonlinear filtering

In this paper, we propose a novel approach to nonlinear filtering utilizing on-line quantization. We develop performance bounds for the algorithm. We also present an example which illustrates the performance of the method.     

Estimation in the multiprocess dynamic generlized linear model

The dynamic generalized linear model and the dynamic discount Bayesian model have been used to describe processes involving time-varying parameters. This paper develops an estimation algorithm for the multiprocess extension of these model. These algorithms have the same characteristics as Harrison-Steven forecasting, namely insensitivity to outliers and quick reaction to real change in the parameters.     

Space–time calibration of radar rainfall data

Motivated by a specific problem concerning the relationship between radar reflectance and rainfall intensity, the paper develops a space–time model for use in environmental monitoring applications. The model is cast as a high dimensional multivariate state space time series model, in which the cross-covariance structure is derived from the spatial context of the component series, in such a way that its interpretation is essentially independent of the particular set of spatial locations at which the data are recorded. We develop algorithms for estimating the parameters of the model by maximum likelihood, and for making spatial predictions of the radar calibration parameters by using realtime computations. We apply the model to data from a weather radar station in Lancashire, England, and demonstrate through empirical validation the predictive performance of the model.     

Prediction and forecasting in linear models with measurement error

We consider measurement error models within the time series unobserved component framework. A variable of interest is observed with some measurement error and modelled as an unobserved component. The forecast and the prediction of this variable given the observed values is given by the Kalman filter and smoother along with their conditional variances. By expressing the forecasts and predictions as weighted averages of the observed values, we investigate the effect of estimation error in the measurement and observation noise variances. We also develop corrected standard errors for prediction and forecasting accounting for the fact that the measurement and observation error variances are estimated by the same sample that is used for forecasting and prediction purposes. We apply the theory to the Yellowstone grizzly bears and US index of production datasets.     

Variance stabilizing filters#

Abstract

In this paper new filters for removing unspecified form of heteroscedasticity are proposed. The filters build on the assumption that the variance of a pre-whitened time series can be viewed as a latent stochastic process by its own. This makes the filters flexible and useful in many situations. A simulation study shows that removing heteroscedasticity before fitting a model leads to efficiency gains and bias reductions when estimating the parameters of ARMA models. A real data study shows that pre-filtering can increase the forecasting precision of quarterly US GDP growth.     

Multivariate mixtures of Erlangs for density estimation under censoring

Multivariate mixtures of Erlang distributions form a versatile, yet analytically tractable, class of distributions making them suitable for multivariate density estimation. We present a flexible and effective fitting procedure for multivariate mixtures of Erlangs, which iteratively uses the EM algorithm, by introducing a computationally efficient initialization and adjustment strategy for the shape parameter vectors. We furthermore extend the EM algorithm for multivariate mixtures of Erlangs to be able to deal with randomly censored and fixed truncated data. The effectiveness of the proposed algorithm is demonstrated on simulated as well as real data sets.     

Multivariate non-linear time series modelling of exposure and risk in road safety research

Summary.  A multivariate non-linear time series model for road safety data is presented. The model is applied in a case-study into the development of a yearly time series of numbers of fatal accidents (inside and outside urban areas) and numbers of kilometres driven by motor vehicles in the Netherlands between 1961 and 2000. The model accounts for missing entries in the disaggregated numbers of kilometres driven although the aggregated numbers are observed throughout. We consider a multivariate non-linear time series model for the analysis of these data. The model consists of dynamic unobserved factors for exposure and risk that are related in a non-linear way to the number of fatal accidents. The multivariate dimension of the model is due to its inclusion of multiple time series for inside and outside urban areas. Approximate maximum likelihood methods based on the extended Kalman filter are utilized for the estimation of unknown parameters. The latent factors are estimated by extended smoothing methods. It is concluded that the salient features of the observed time series are captured by the model in a satisfactory way.     

Estimating multivariate autoregressive moving average models by fitting long autoregressions

Some simple methods for the estimation of mixed multivariate autoregressive moving average time series models are introduced. The methods require the fitting of a long autoregression to the data and the computation of consistent initial estimates for the parameters of the model. After these preliminaries the estimators of the paper are obtained by applying weighted least squares to a multivariate auxiliary regression model. Two types of weight matrices are considered. Both of them yield estimators which are strongly consistent and asymptotically normally distributed. The first estimators are also asymptotically efficient while the second ones are not fully efficient but computationally simple. A simulation study is performed to illustrate the behaviour of the estimators in finite samples.