Replicated INAR(1) Processes

Replicated time series are a particular type of repeated measures, which consist of time-sequences of measurements taken from several subjects (experimental units). We consider independent replications of count time series that are modelled by first-order integer-valued autoregressive processes, INAR(1). In this work, we propose several estimation methods using the classical and the Bayesian approaches and both in time and frequency domains. Furthermore, we study the asymptotic properties of the estimators. The methods are illustrated and their performance is compared in a simulation study. Finally, the methods are applied to a set of observations concerning sunspot data. AMS 2000 Subject Classification 62M10, 91B70, 60G10     

Modelling with the Novel INAR(1)-PTE Process

Methodology and Computing in Applied Probability - In this paper, the first-order non-negative integer-valued autoregressive process with Poisson-transmuted exponential innovations is introduced....     

Inferential aspects of the zero-inflated Poisson INAR(1) process

A first-order INteger-valued AutoRegressive (INAR) process with zero-inflated Poisson distributed innovations was proposed by Jazi, Jones and Lai (2012) [First-order integer valued AR processes with zero inflated Poisson innovations. Journal of Time Series Analysis. 33, 954–963.], which is able for dealing with zero-inflated/deflated count time series data. The inferential aspects of this model were not well explored by the authors, only a conditional maximum likelihood approach was briefly discussed. In this paper, we explore the inferential aspects of this zero-inflated Poisson INAR(1) process. We propose parameter estimation through Two-Step Conditional Least Squares and Yule–Walker methods. The asymptotic properties of the estimators are provided. Simulation results about the finite-sample behavior of both estimation methods and comparisons with the conditional maximum likelihood approach are presented under correct model specification and misspecification. Two empirical applications to real data sets are considered in order to illustrate the usefulness of the proposed methodology in practical situations.     

On modifications of random processes

Translated fromZapiski Nauchnykh Seminarov POMI, Vol. 194, pp. 150–169, 1992.     

A bivariate INAR(1) time series model with geometric marginals

In this paper we introduce a simple bivariate integer-valued time series model with positively correlated geometric marginals based on the negative binomial thinning mechanism. Some properties of the model are considered. The unknown parameters of the model are estimated using the modified conditional least squares method.     

Risk aggregation based on the Poisson INAR(1) process with periodic structure

In this paper, we consider a risk model by introducing a temporal dependence between the claim numbers under periodic environment, which generalizes several discrete-time risk models. The model proposed is based on the Poisson INAR(1) process with periodic structure. We study the moment-generating function of the aggregate claims. The distribution of the aggregate claims is discussed when the individual claim size is exponentially distributed.     

Quantile Regression for Thinning-based INAR(1) Models of Time Series of Counts

In this paper,we develop the quantile regression(QR)estimation for the first-order integer-valued autoregressive(INAR(1))models by defining the smoothing INAR(1)process.Jittering method is used to derive the QR estimators for the autoregressive coefficient and the quantile of innovations.The consistency and asymptotic normality of the proposed estimators are established.The performances of the proposed estimation procedures are evaluated by Monte Carlo simulations.The results show that the proposed procedures perform well for simulations and a real data application.     

On random observation processes for stochastic approximation

Summary In [1] a new procedure is given to estimate the root of a regression equation. The purposes of this paper are to extend the Lemma 1 in [1] and to give a process involving a randomly determined sequence of observations for finding thermaximum of a regression function. The process is similar to that of [1]. Kiefer and Wolfowitz [2] gave a stochastic approximation procedure for the latter purpose. Their process needs the condition of the unimodality of the regression function which is not required for our case.     

Some geometric mixed integer-valued autoregressive (INAR) models

In this paper, we introduce some mixed integer-valued autoregressive models of orders 1 and 2 with geometric marginal distributions, denoted by MGINAR(1) and MGINAR(2), using a mixture of the well-known binomial and the negative binomial thinning. The distributions of the innovation processes are derived and several properties of the model are discussed. Conditional least squares and Yule-Walker estimators are obtained, and some numerical results of the estimations are presented. A real-life data example is investigated to assess the performance of the models.     

Weak limits of random coefficient autoregressive processes and their application in ruin theory

We prove that a large class of discrete-time insurance surplus processes converge weakly to a generalized Ornstein–Uhlenbeck process, under a suitable re-normalization and when the time-step goes to 0. Motivated by ruin theory, we use this result to obtain approximations for the moments, the ultimate ruin probability and the discounted penalty function of the discrete-time process.     

Random flows generated by random processes (properties of random flows,level-crossing problems)

Author's summary of a dissertation submitted for the degree of Doctor of Physicomathematical Sciences. The dissertation was defended February 10, 1970, at a meeting of the Scientific Council of the Institute of Applied Mathematics, Academy of Sciences of the USSR. Official opponents: B. V. Gnedenko, Academician of the Academy of Sciences of the USSR, Yu. V. Prokhorov, Corresponding Member of the Academy of Sciences of the USSR, and N. N. Chentsov, Doctor of Physicomathematical Sciences.Translated from Matematicheskie Zametki, Vol. 8, No. 3, pp. 393–407, September, 1970.     

On infinitesimal σ-fields generated by random processes

It is proved that the infinitesimal look-ahead and look-back σ-fields of a random process disagree at atmost countably many time instants.     

On first and second order stationarity of random coefficient models

We give conditions for first and second order stationarity of mixture autoregressive processes. We obtain a simple condition for positive definiteness of the solution of a generalisation of the Stein’s equation with semidefinite right-hand side and apply it to second order stationarity. The said condition may be of independent interest.     

On the safety stock problem for random delivery processes

Inventory models are considered in which the delivery of an order occurs not on one occasion but at random moments of a period in random parts. We give two extensions of the reliability type inventory model of A. Prékopa. In this model a known constant demand rate is assumed and a simple approximate formula is given for the initial stock of the order period which serves as safety stock and ensures a continuous supply during the whole order period on a prescribed probability level. This formula is widely used in practice for safety stock planning in the case when deliveries in random parts occur.We formulate a generalized version of the random delivery process and derive the exact solution of the safety stock which can be applied also for the previous model. In the second model a random demand rate is considered together with a random delivery process. An exact solution method and a simple approximate formula for the safety stock will be discussed. We have experiences in the application of these models both in a steel works and a textile factory in Hungary.     

On small deviation probabilities for certain iterated random processes

The asymptotic behavior of small deviation probabilities for some iterated random processes is investigated. It is shown that, under certain conditions, iterated and noniterated processes have logarithmic asymptotics of the same character; otherwise, these asymptotics may differ substantially. Some iterated Gaussian processes are considered as an example.