營利事業所得稅資料之隨機模型

Abstract

In business income tax computation the items list in the tax return form are mostly recorded at consecutive equally-spaced time instants. Linear stochastic models are found to be appropriate to fit these items.Firstly, the data has to be converted into a proper form suitable for fitting. Then a procedure is used to find a model of the series.It involves the model identification,parameter estimation and model diagnostic evaluation.Finally, the model obtained is used to predict the future values of the series based on a minimum mean square error criterion.