文献类型：会议论文

英文题名：College Enrollment Forecasts Using Artificial Intelligence and Time Series Models

作者：Chen, Chau-Kuang[1];Yang, Aiping[2]

第一作者：Chen, Chau-Kuang

通讯作者：Chen, C.-K.

机构：[1]Meharry Med Coll, Off Inst Res, Nashville, TN 37208 USA;[2]Beijing Union Univ, Dept Ind Engn, Beijing 100020, Peoples R China

第一机构：Meharry Med Coll, Off Inst Res, Nashville, TN 37208 USA

会议论文集：15th World Multi-Conference on Systematics, Cybernetics and Informatics (WMSCI 2011)

会议日期：JUL 19-22, 2011

会议地点：Orlando, FL

语种：英文

外文关键词：Enrollment Forecast; ANN; SVM; GEP; ARIMA

摘要：Decision-makers in colleges and universities need high quality enrollment forecasts to appropriately establish proper resources for academic programs and support services. However, accurate forecasts are difficult to make due to some fluctuation in the enrollment from year to year. Also, certain important factors affecting student enrollment are difficult to quantify. In addition, various forecasting techniques and related software packages may add to the technical complexity. In this study, ANN, SVM, and GEP modeling approaches were used to perform enrollment forecasts for Oklahoma State University from 1962 to 2010. The ARIMA model was also built as a benchmarking tool to verify model accuracy. Nine independent variables were entered into the model equations in an attempt to increase explanatory power. These variables include Oklahoma high school graduates, competitor college enrollment from the University of Oklahoma, state funding, and economic indicators such as state unemployment rate, gross national product, and consumer price index. The empirical results indicate that ANN and SVM models yielded remarkable model fitting statistic and exceptionally small forecasting error. ANN and SVM models have demonstrated their model validity and accuracy. Hence, they could be replicated for comparable universities elsewhere.