摘要:

AbstractThis paper investigates the performance of Artificial Neural Networks for the classification and subsequent prediction of business entities into failed and non‐failed classes. Two techniques, back‐propagation and Optimal Estimation Theory (OET), are used to train the neural networks to predict bankruptcy filings. The data are drawn from Compustat data tapes representing a cross‐section of industries. The results obtained with the neural networks are compared with other well‐known bankruptcy prediction techniques such as discriminant analysis, probit and logit, as well as against benchmarks provided by directly applying the bankruptcy prediction models developed by Altman (1968) and Ohlson (1980) to our data set. We control the degree of ‘disproportionate sampling’ by creating ‘training’ and ‘testing’ populations with proportions of bankrupt firms ranging from 1% to 50%. For each population, we apply each technique 50 times to determine stable accuracy rates in terms of Type I, Type II and Total Error. We show that the performance of various classification techniques, in terms of their classification errors, depends on the proportions of bankrupt firms in the training and testing data sets, the variables used in the models, and assumptions about the relative costs of Type I and Type II errors. The neural network solutions do not achieve the ‘magical’ results that literature in this field often promises, although there are notable 'pockets' of superior performance by the neural networks, depending on particular combinations of proportions of bankrupt firms in training and testing data sets and assumptions about the relative costs of Type I and Type II errors. However, since we tested only one architecture for the neural network, it will be necessary to investigate potential improvements in neural network performance through systematic changes in neural network architecture.

ISSN:1055-615X    

DOI:10.1002/j.1099-1174.1995.tb00083.x

出版商:Wiley    

年代:2014

数据来源: WILEY

摘要:

AbstractThe detection of management fraud is an important issue facing the auditing profession. A major contributor to this issue is the Loebbecke and Willingham (1988) conceptual model for the detection of management fraud. A cascaded Logit approach using the Loebbecke and Willingham model was developed in Bellet al.(1993). The present study offers an alternative approach using Artificial Neural Networks (ANNs). This paper develops a successful discriminator of management fraud using both the generalized adaptive neural network architectures (GANNA) and the Adaptive Logic Network (ALN) approaches to designing neural networks. The discriminant functions can distinguish between fraudulent and non‐fraudulent companies with superior accuracy to the cascaded Logit results of Bellet al.(1993). Finally, the discriminant function provides a parsimonious set of questions useful for detecting management fraud.

ISSN:1055-615X    

DOI:10.1002/j.1099-1174.1995.tb00084.x

出版商:Wiley    

年代:2014

数据来源: WILEY

摘要:

AbstractThere exists today an unanswered question as to whether, and the degree to which, borrower characteristics impact the choice between fixed and adjustable rate mortgages. In this paper, we apply a neural network analysis to supply evidence that answers this question. We find evidence that the characteristics of a borrower's net worth, marital status and education level and whether a co‐borrower is involved contribute in a significant way to the neural network's ability to determine mortgage choice. Further, we show how, because of the facility of neural networks in modeling intrasample differences, they achieve material and statistically significant accuracy gains over qualitative choice models in predicting whether a borrower will choose a fixed or adjustable rate mortgage.

ISSN:1055-615X    

DOI:10.1002/j.1099-1174.1995.tb00085.x

出版商:Wiley    

年代:2014

数据来源: WILEY

摘要:

AbstractDecision‐support systems can be improved by enabling them to use past decisions to assist in making present ones. Reasoning from relevant past cases is appealing because it corresponds to some of the processes an expert uses to solve problems quickly and accurately. All this depends on an effective method of organizing cases for retrieval. This paper investigates the use of inductive networks as a means for case organization and outlines an approach to determining the desired number of cases—or assessing the reliability of a given number. Our method is demonstrated by application to decision making on corporate tax audits.

ISSN:1055-615X    

DOI:10.1002/j.1099-1174.1995.tb00086.x

出版商:Wiley    

年代:2014

数据来源: WILEY

ISSN:1055-615X    

DOI:10.1002/j.1099-1174.1995.tb00087.x

出版商:Wiley    

年代:2014

数据来源: WILEY