Methods for Multi-Step Time Series Forecasting Neural Networks

In this study, we examine two methods for Multi-Step forecasting with neural networks: the Joint Method and the Independent Method. A subset of the M-3 Competition quarterly data series is used for the study. The methods are compared to each other, to a neural network Iterative Method, and to a baseline de-trended de-seasonalized naïve forecast. The operating characteristics of the three methods are also examined. Our findings suggest that for longer forecast horizons the Joint Method performs better, while for short forecast horizons the Independent Method performs better. In addition, the Independent Method always performed at least as well as or better than the baseline naïve and neural network Iterative Methods.

Forecasting Stock Returns with Artificial Neural Networks

During the last few years there has been growing literature on applications of artificial neural networks to business and financial domains. In fact, a great deal of attention has been placed in the area of stock return forecasting. This is due to the fact that once artificial neural network applications are successful, monetary rewards will be substantial. Many studies have reported promising results in successfully applying various types of artificial neural network architectures for predicting stock returns. This chapter reviews and discusses various neural network research methodologies used in 45 journal articles that attempted to forecast stock returns. Modeling techniques and suggestions from the literature are also compiled and addressed. The results show that artificial neural networks are an emerging and promising computational technology that will continue to be a challenging tool for future research.

Forecasting Emerging Market Indexes with Neural Networks

Forecasting financial time series with neural networks is problematic. Multiple decisions, each of which affects the performance of the neural network forecasting model, must be made, including which data to use and the size and architecture of the neural network system. While most previous research with neural networks has focused on homogenous models, that is, only using data from the single time series to be forecast, the ever more global nature of the world’s financial markets necessitates the inclusion of more global knowledge into neural network design. This chapter demonstrates how specific markets are at least partially dependent on other global markets and that inclusion of heterogeneous market information will improve neural network forecasting performance over similar homogeneous models by as much as 12 percent (i.e., moving from a near 51% prediction accuracy for the direction of the market index change to a 63% accuracy of predicting the direction of the market index change).

Using Artificial Neural Networks to Forecast Market Response

Marketing managers must quantify the effects of marketing actions on contemporaneous and future sales performance. This chapter examines forecasting with artificial neural networks in the context of model-based planning and forecasting. The emphasis here is on causal modeling; that is, forecasting the impact of marketing mix variables, such as price and advertising, on sales.

Assessment of Evaluation Methods for Prediction and Classifications of Consumer Risk in the Credit Industry

In this chapter, we examine and compare the most prevalent modeling techniques in the credit industry, Linear Discriminant Analysis, Logistic Analysis and the emerging technique of Neural Network modeling. K-S Tests and Classification Rates are typically used in the industry to measure the success in predictive classification. We examine those two methods and a third, ROC Curves, to determine if the method of evaluation has an influence on the perceived performance of the modeling technique. We found that each modeling technique has its own strengths, and a determination of the “best” depends upon the evaluation method utilized and the costs associated with misclassification.

A Weighted Window Approach to Neural Network Time Series Forecasting

In this chapter, we propose a neural network based weighted window approach to time series forecasting. We compare the weighted window approach with two commonly used methods of rolling and moving windows in modeling time series. Seven economic data sets are used to compare the performance of these three data windowing methods on observed forecast errors. We find that the proposed approach can improve forecasting performance over traditional approaches.

Predicting Consumer Situational Choice with Neural Networks

This study shows how neural networks can be used to model posterior probabilities of consumer choice and a backward elimination procedure can be implemented for feature selection in neural networks. Two separate samples of consumer choice situations were selected from a large consumer panel maintained by AT&T. Our findings support the appropriateness of using neural networks for these two purposes.

Forecasting Short-Term Exchange Rates

Movements in foreign exchange rates are the results of collective human decisions, which are the results of the dynamics of their neurons. In this chapter, we demonstrate how to model these types of market behaviors by recurrent neural networks (RNN). The RNN approach can help us to forecast the short-term trend of foreign exchange rates. The application of forecasting techniques in the foreign exchange markets has become an important task in financial strategy. Our empirical results show that a discrete-time RNN performs better than the traditional methods in forecasting short-term foreign exchange rates.

Business Forecasting with Artificial Neural Networks

Artificial neural networks have emerged as an important quantitative modeling tool for business forecasting. This chapter provides an overview of forecasting with neural networks. We provide a brief description of neural networks, their advantages over traditional forecasting models, and their applications for business forecasting. In addition, we address several important modeling issues for forecasting applications.

Using an Extended Self-Organizing Map Network to Forecast Market Segment Membership

This chapter presents an extended Self-Organizing Map (SOM) network and demonstrates how it can be used to forecast market segment membership. The Kohonen’s SOM network is an unsupervised learning neural network that maps n-dimensional input data to a lower dimensional (usually one- or two-dimensional) output map while maintaining the original topological relations. We apply an extended version of SOM networks that further groups the nodes on the output map into a user-specified number of clusters to a residential market data set from AT&T. Specifically, the extended SOM is used to group survey respondents using their attitudes towards modes of communication. We then compare the extended SOM network solutions with a two-step procedure that uses the factor scores from factor analysis as inputs to K-means cluster analysis. Results using AT&T data indicate that the extended SOM network performs better than the two-step procedure.