Neural networks have been shown to be a promising tool for forecasting financial time series. Several design factors significantly impact the accuracy of neural network forecasts. These factors include selection of input variables, architecture of the network, and quantity of training data. The questions of input variable selection and system architecture design have been widely researched, but the corresponding question of how much information to use in producing high-quality neural network models has not been adequately addressed. In this paper, the effects of different sizes of training sample sets on forecasting currency exchange rates are examined. It is shown that those neural networks---given an appropriate amount of historical knowledge ---can forecast future currency exchange rates with 60 percent accuracy, while those neural networks trained on a larger training set have a worse forecasting performance. In addition to higher-quality forecasts, the reduced training set sizes reduce development cost and time.

Artificial neural networks have been used to support applications across a variety of business and scientific disciplines during the past years. Artificial neural network applications are frequently viewed as black boxes which mystically determine complex patterns in data. Contrary to this popular view, neural network designers typically perform extensive knowledge engineering and incorporate a significant amount of domain knowledge into artificial neural networks. This paper details heuristics that utilize domain knowledge to produce an artificial neural network with optimal output performance. The effect of using the heuristics on neural network performance is illustrated by examining several applied artificial neural network systems. Identification of an optimal performance artificial neural network requires that a full factorial design with respect to the quantity of input nodes, hidden nodes, hidden layers, and learning algorithm be performed. The heuristic methods discussed in this paper produce optimal or near-optimal performance artificial neural networks using only a fraction of the time needed for a full factorial design. Keywords: Artificial neural networks; Heuristics; Input vector; Hidden layer size; ANN learning method; Design. Heuristics Principles for the Design of Artificial Neural Networks - Page 3 1.

This article describes the use of neural networks as an alternative method to investigate the links between various dimensions of culture and perceptions of justice and demonstrates their ability to model the data relationships with higher accuracy than multiple regression analysis. A complete discussion of the development and validation of the neural network models is included as a guide to researchers in management who are interested in exploring this methodology.

Today most organizations run their daily operations using data at their disposal. However, a vast majority of the organizations do not have adequate process and tools to maintain high quality operational data at all times. One of the key reasons for this is the lack of appreciation of the damages that low quality data can bring to an organization, and the cost of ensuring high quality of data. This article provides a basis for quantifying in monetary terms the costs of both low quality data and ensuring high quality data. A comparison of the costs of low quality data and ensuring high quality data can be a simple and compelling basis for an organization to determine the extent of the efforts it must expend to ensure high quality of its operational data. 1 TYPES OF OPERATIONAL DATA To provide a basis for understanding the costs of both low quality data and ensuring high quality data, we need to understand the types of data that organizations use in running their daily operations. There are at least five types: • “Front-office ” data

Although databases used in many organizations have been found to contain errors, little is known about the effect of these errors on predictions made by linear regression models. The paper uses a real-world example, the prediction of the net asset values of mutual funds, to investigate the effect of data quality on linear regression models. The results of two experiments are reported. The first experiment shows that the error rate and magnitude of error in data used in model prediction negatively affect the predictive accuracy of linear regression models. The second experiment shows that the error rate and the magnitude of error in data used to build the model positively affect the predictive accuracy of linear regression models. All findings are statistically significant. The findings have managerial implications for users and builders of linear regression models.

An application of neural networks to classify the performance status of business firms is performed. An artificial neural network (ANN) model is developed using publicly available dataset as input and output variables. Several different neural network topologies are designed and applied to the datasets. A neural network model classifies both high and low business performance status. The ANN model can enhance strategic and managerial insights by providing meaningful financial information.

Multilayer neural network has been successfully applied to the time series forecasting. Steepest descend, a popular learning algorithm for backpropagation network, converges slowly and has the difficulty in determining the network parameters. In this paper, conjugate gradient learning algorithm with restart procedure is introduced to overcome these problems. Also, the commonly used random weight initialization does not guarantee to generate a set of initial connection weights close to the optimal weights leading to slow convergence. Multiple linear regression (MLR) provides a better alternative for weight initialization. The daily trade data of the listed companies from Shanghai Stock Exchange is collected for technical analysis with the means of neural networks. Two learning algorithms and two weight initializations are compared. The results find that neural networks can model the time series satisfactorily, whatever which learning algorithm and weight initialization are adopted. However, the proposed conjugate gradient with MLR weight initialization requires a lower computation cost and learns better than steepest decent with random initialization.