Simple models for positive-valued and discrete-valued time series with ARMA correlation structure

摘要: Abstract : Three models for positive-valued and discrete-valued stationary time series are discussed. All have the property that a range of specified marginal distributions same correlation structure as usual linear, autoregressive-moving average (ARMA) model. The differ in which can be accommodated simplicity flexibility each Specifically EARMA-type processes extended from exponential distribution to rather narrow continuous distributions; DARMA-type defined usefully any discrete simple flexible. Finally marginally controlled semiMarkov generated process or is therefore very However, model suffers some complexity parametric obscurity. (Author)