Neural Networks for Time Series Prediction: Practical Implications of Theoretical Results

• Activity Number: 299
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Melinda F. Thielbar*+ and David A. Dickey
• Companies: North Carolina State University and North Carolina State University
• Keywords: time series ; nonlinear time series ; neural networks ; time series forecasting ; forecasting

Abstract

Research on the performance of neural networks in modeling nonlinear time series has produced mixed results. We propose adding to the existing literature by examining theoretical results (Tong 1993, Leoni 2009, Trapletti 2000) and using them as a guide for developing best practices in neural network modeling. Using simulations, we place conflicting empirical results (Adaya and Collopy 1998, Nam and Schaefer 1995, Faraway and Chatfield 1997, Taskaya-Temizel and Casy 2005) in context and examine their implications for future research.