Comparison of Two Time Series Decomposition Models of Quadratic Trend-Cycle Components
Kelechukwu C. N. Dozie *
Department of Statistics, Imo State University, Owerri, Imo State, Nigeria.
Stephen O. Ihekuna
Department of Statistics, Imo State University, Owerri, Imo State, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
This paper presents comparison of two time series decomposition models of quadratic trend component with emphasis on the mixed and multiplicative models. Therefore, the ultimate is to compare the row, column and overall averages and variances of the series that admit the mixed and multiplicative models The method adopted in obtaining the estimators of the parameters are those proposed for both mixed and multiplicative models. Results indicate that (1) the seasonal means of the mixed model is a quadratic multiple of the square of the seasonal effect. The quadratic is in j. It depends on the seasonal component (\(S^2_j\)) of the jth column for the multiplicative model (2) for the mixed model, the error variances assumed equal to one. The error variances of multiplicative model are multiplied (3) the variations between the periodic means and standard deviations of the mixed model is more than that of the multiplicative model.
Keywords: Decomposition model, quadratic curve, trend parameter, seasonal indices, buys-ballot table.