Water Level Data Modeling with Bilinear Time Series Analysis

Abstract

In the literature, many time series data, such as the economic and hydrological data, show various nonlinearity characteristics. The Keenan's test and F-test are employed in identifying a nonlinear data set. This article looks at the modeling of nonlinear time series data using bilinear time series model. The model is an extension of autoregressive model such that an extra term representing the bilinear characteristic is introduced. The estimation of bilinear models is obtained using nonlinear least squares method. As an illustration, analysis on water level of Sungai Kelantan using the above method is presented.