COMPUTER MODELS FOR ENROLLMENT FORECASTING: A MANAGEMENT SCIENCE APPROACH

Forecasting student enrollment trends is frequently accompanied by considerable uncertainty which usually leads to unwieldly problems in determining school facility needs. However, highly sophisticated computerized models can be effectively used to reduce the elements of risk and uncertainty in enrollment forecasting.; The purpose of this study was to build three computer models and to test their effectiveness in forecasting student enrollment in educational systems.; The major elements of the study were: (1) To construct the following management science models which can be used in enrollment forecasting: (a) The linear regression model. (b) The moving average model. (c) The exponential model: (i) The exponential smoothing sub-model. (ii) The binomial sub-model. (iii) The trinomial sub-model. (2) To build the corresponding computer programs for the above management science models. (3) To test the computer models by applying them retrospectively to a micro or macro educational system.; The null hypotheses of the study were: (1) There is no statistically significant difference between the population mean of the forecasts obtained by the Linear Regression Computer Model (LRCM); that obtained by the Moving Average Computer Model (MACM); and that obtained by the Exponential Computer Model (ECM). (2) There is no statistically significant difference between the population mean of the forecasts obtained by LRCM and that obtained by MACM. (3) There is no statistically significant difference between the population mean of the forecasts obtained by LRCM and that obtained by ECM. (4) There is no statistically significant difference between the population mean of the forecasts obtained by MACM and that obtained by ECM.; The research method was model building and hypothesis testing.; The major findings of the study were: (1) The first null hypothesis was rejected at the 0.00001 level of significance. (2) The second null hypothesis was rejected at the 0.001 level of significance. (3) The third null hypothesis was not rejected. (4) The fourth null hypothesis was rejected at the 0.010 level of significance. (5) The difference between the mean of the Actual Transition Numbers of students enrolled in the forecasted period (ATN) and the mean of LRCM is statistically significant at the 0.00028 level of significance. (6) The difference between the mean of ATN and the mean of MACM is statistically significant at the 0.034 level of significance. (7) The difference between the mean of ATN and the mean of ECM is not statistically significant.; The results of the study indicate that the forecasts obtained through the application of ECM provide more accurate estimates than the forecasts obtained through the applications of both LRCM and MACM. The results of the study also indicate that the forecasts obtained through the application of LRCM provide more accurate estimates than the forecasts obtained through the application of MACM.