Using Transfer Function Models to Build a Statistical Model to Forecasting Student Graduation Rates

Author:   Obaid Mahmmood Mohsin Alzawbaee and Nozad H. Mahmood

Download PDF

Abstract

Transfer function model is one of the quantitative forecasting models that can be used to forecast multivariate time series, and it can be used to describe the dynamic effect of the input series {X_t}  on the output series {Y_t}  , as well as it can be used in forecasting .

We use this technique to build a model by which student's   graduation rates can be forecasted based on two series data: the average rate of the student for the first, second, and third years, which represents the input series (X_t) (the weighted rate), and the graduation rate of the student, which represents the output series (Y_t). The size of the series is  (n = 114),we used (111) of which , and (3) were left for forecasting purposes.

The most important of conclusions and recommendations are: The bivariate time series  are stationary . The two series were prewhitened and it became clear that the appropriate model for both series is ARMA (1, 1) model. We built appropriate transfer function model which gives accuracy in forecasting.

Keywords: multivariate time series, Transfer function model, forecasting, ARMA models.